Patent Practitioners’ Changing Role in Drafting Documents

Share this article

By Ian C. Schick, Ph.D., Esq. (first published in ABA Landslide Magazine)

    Ian C. Schick is founder and CEO of Draft Builders and Specifio, fellow at Stanford Law School’s CodeX Center for Legal Informatics, and chair of AIPLA’s Emerging Technologies Committee.

The manner in which patent documents are created has always been evolving—from handwritten patent applications to typescript, from typewriters to word processors, from dictation to auto-transcription, examples abound. The drive for better quality work product, enhanced efficiency, and improved work experience for practitioners has made this a continuous process since the beginning of the modern patent system.

Patent Document Preparation and Artificial Intelligence

In recent years, artificial intelligence (AI) and the automation it brings have taken on an increasing importance in preparing patent documents.[1] Patent-specific, automated proofreaders, for example, which have now largely replaced the need for manual checking for minor informalities, first appeared over 15 years ago.[2] The first “dynamic document” patent editor (e.g., if a practitioner changes a label in a figure, corresponding text in the specification automatically changes accordingly) hit the market almost 10 years ago.[3] In the last five years, automated patent content generation (APCG or “auto-drafting”) has become commercially available.[4]

Current technologies in APCG are surprisingly precise but offer relatively limited solutions, typically focusing on only certain parts of a patent document and/or on only certain technology fields. However, unlike proofreading for informalities and synchronizing figure labels, the generation of content for patent documents hits at the heart of practitioners’ unique skills and value proposition. For this reason, and perhaps unsurprisingly, APCG has been met with frequent skepticism over efficacy and flat-out fear about job displacement.[5] Allaying these anxieties starts with a better understanding of where APCG stands today, how its development will likely progress into the future, and what that means for patent practitioners’ ever-evolving role in drafting documents.

In contemporary parlance, AI is essentially synonymous with automation for most contexts. AI is generally understood to describe “[a]n algorithm or machine capable of completing tasks that would otherwise require cognition.”[6] AI comes in three flavors: narrow AI (or weak AI), general AI (or strong AI), and super AI. Of these categories, it is only narrow AI that exists today and, by most accounts, is the only type of AI that will exist for the foreseeable future.[7]

Narrow AI describes a computer program that is good at performing a defined set of tasks (e.g., tasks associated with playing chess or Go or making purchase suggestions, sales predictions, or weather forecasts). In the broadest sense, today’s AI includes nonlearning systems that automate traditional human tasks (e.g., rule-based expert systems). Machine learning is a subset of current AI where hard coded algorithms are replaced by models trained on example input-output pairs to predict outputs for previously unseen inputs. Deep learning is a subset of machine learning that employs vast networks of artificial neurons. General AI is a purely hypothetical computer program that can understand and reason its environment as a human would. Also purely hypothetical, super AI describes a computer program that is much smarter than the sum of all human intelligence in practically every field.

Impact of Automation on Other Professional Services

Given the circumscribed capabilities of existing AI-based automation as well as the subtle nuance and broad contextual understanding required in drafting patent documents, the chances seem quite remote that AI will completely displace practitioners anytime soon. In fact, in many professional services industries, automation has led to more professionals rather than fewer. Take, for example, electronic spreadsheets in accounting[8] and computer-aided drafting in architecture.[9] In both cases, these disruptive technologies resulted in “a net positive for the industry with higher quality, higher efficiency, better access to services, and growth in the workforce.”[10] Will patent practitioners see the same in their industry with advances in APCG?

Motivation for Advancement in Automated Patent Content Generation

The motivating factors for advancement in APCG are straightforward and stem from a pronounced nonequilibrium in the U.S. patent marketplace. The last several decades have seen sustained growth in demand for patent services, with patent application filings and substantive official actions mailed up about 30% and 40%, respectively, over a recent decade.[11] Supply, in the form of active patent practitioners, has largely collapsed over the last 10 years, particularly in early-career practitioners who historically have served as important laboring oars in patent document drafting.[12] With rising demand and shrinking supply, a market at equilibrium would see rising prices. The reality, however, is a long-term downward trend in practitioner fees for preparing patent applications and office action responses.[13] For example, inflation-adjusted average fees to prepare and file a software patent application have decreased by about 34% over the past decade.

In the demand-supply-price equation, supply is the only thing the patent preparation and prosecution industry itself can affect. Since more patent practitioners will not arrive overnight, increasing supply means increasing per-practitioner document production without relying on adding significantly more practitioners. Efforts around this traditionally included using nonattorney practitioners (i.e., patent agents) and nonlicensed technical writers (e.g., patent engineers, technical specialists, etc.) for drafting work. Simply swapping out patent attorneys for patent agents or patent agents for patent engineers, however, is not a sustainable approach because it just taps another limited talent pool and maintains human inefficiencies (i.e., time, cost, and errors), but with lesser-trained individuals.

Patent Documents and Lean Production

A better approach to generating more substantive patent documents per patent practitioner requires treating document production as a manufacturing process that can realize the benefits of lean production principles.[14] In effect, a patent document preparation process should be viewed as a series of separate but interlinked subprocesses, with each subprocess being delegated to the most efficient (i.e., least expensive) resource without sacrificing work product quality or increasing chances for errors. If possible, each subprocess should be delegated to a computer (i.e., automated). If automation is not possible, the subprocess should be delegated to a “less expensive” human resource (e.g., nonexpert versus expert). And if that is not possible, the subprocess should not be delegated and, instead, be performed by a patent practitioner (i.e., the most expensive resource).

Content Type as a Framework to Analyze Roles in Document Production

Without having to enumerate all potential subprocesses in drafting a patent document, the parts that can be delegated to a computer can be identified by analyzing the different types of content that exist in all patent documents; namely, bespoke writing content, mechanical writing content, and canned text.[15]

Generating bespoke writing content is where patent practitioners provide their primary value-add. This content reflects the intellectual heavy lifting performed by the practitioner preparing the document. It often involves original analysis on unique facts and is driven by creativity, judgment, strategy, experience, and contextual knowledge about the project at hand (e.g., assignee’s business objectives, known prior art, competitor activity, etc.). Bespoke writing content is too nuanced, context dependent, and consequential for wholesale automation. Examples of bespoke writing content include patent claims, problem/solution statements, description of key concepts and nuances of the invention, background section, and, for prosecution, claim amendments and arguments.

Mechanical writing content represents the rote and mundane parts of traditional writing projects. This content is driven by convention and/or by satisfying document requirements. It must be accurate and complete but does not require significant mental work. Traditionally, generating mechanical writing content includes a manual “copy, paste, massage” of claim language or text from a separate resource. For example, generating mechanical writing content includes propagating claim language throughout the specification (e.g., title field of invention, summary, literal claim support in detailed description, additional claim sets mirroring attorney-written claims, abstract, and method flow charts). It also generally includes manually extracting information from separate resources such as lists of well-known examples, dictionary definitions, and descriptions of well-known facts.

All patent documents contain some amount of canned text. Canned text is “predetermined language,” such as boilerplate, stock definitions and descriptions, and other reused content. It may be described as “static content” in that it is shared across multiple patent documents.

With content divided by type, it becomes readily apparent where opportunities lie for technical innovation and how practitioners’ role in patent document production will likely evolve and—some would say—improve for the benefit of the entire patent ecosystem (see fig. 1).

Figure 1

Generating Bespoke Writing Content

Unless and until general AI is achieved, generating bespoke writing content will remain the purview of skilled practitioners. At present, this content is entirely human-written and unique for each patent document. Technical advances in generating bespoke writing content will likely focus on accelerating practitioners’ ability to write, and not on replacing practitioners altogether as content generators. For example, predictive text (e.g., akin to what currently exists in Google’s Gmail message editor) may be implemented to provide suggestions for sentence completions, dependent claims, etc. In any case, however, human practitioners will remain central to generating bespoke writing content for some time to come.

Generating Mechanical Writing Content

Advances in auto-generated mechanical writing content have created substantial buzz in recent years, specifically around generating “claim support” for the specification based on human-written claims.[16] These tools, developed at both technology companies and private law firms, are accurate, instant, and reliable. They are accurate in the sense that they convert claim language into corresponding complete sentences that technically do the job of providing literal claim support. However, the auto-generated content can sound stilted and robotic when read.

Improving the readability of auto-generated claim support will accelerate adoption, but getting there will require advancements in claim language transduction and variety in surface realization. In other words, many, many sentence patterns should be used when generating text rather than only a few, as is currently the practice in commercially available systems. Take, for example, input claim language reciting:

    wherein the shaft is made of a material including one or more of iron, steel, aluminum, or copper.”

Contemporary APCG systems might convert the example input above into a sentence that reads:

    By way of nonlimiting example, the shaft may be made of a material including one or more of iron, steel, aluminum, or copper.”

Two key transductions occurred in generating the above output. The input language was converted to permissive prose (i.e., by use of “may be” rather than “is”), and the closed-ended list of the input claim language was converted to an open-ended list (i.e., by use of “[b]y way of nonlimiting example”). The same output sentence pattern, however, would generally be used in existing systems for any input having the same structure of the example input above.

A more advanced system might generate many potential outputs based on many different sentence patterns and suggest one at random or based on user preferences or other factors.[17] Continuing the illustration above, examples of such multiple potential outputs could include:

    “The shaft may be made of a material. By way of nonlimiting example, the material may include one or more of iron, steel, aluminum, or copper”;

    “The shaft material may include iron, steel, aluminum, copper, and/or other materials”; or

    “In some implementations, a material from which the shaft is made may comprise at least one of iron, steel, aluminum, copper, and/or other materials.”

Each of the above outputs reflects the same two transductions (i.e., permissive prose and open-ended lists) and is technically sufficient to provide the desired claim support, but, clearly, varying cadence and sentence structure should give a more natural sound to the reader. Auto-generation of claim support based on claim language has largely been solved in terms of efficacy, and readability will only improve. As such, fewer and fewer practitioners will spend time manually generating this kind of mechanical writing content, with client expectations following suit.

Aside from claim support, generating mechanical writing content includes extracting information from separate resources and massaging it into the document being prepared. Presently, this is a completely manual process typically performed by patent practitioners using dictionaries, technical references, encyclopedias, a law firm’s own past work product, etc. Future systems may automate extraction of definitions and descriptions of examples and well-known facts while respecting any copyright restrictions or requirements. For example, a custom (and perhaps automatically) built “dictionary” with topic-description pairs may be used to automate content generation, or at least accelerate it via suggested content. Content may be pulled from licensed resources, open-source dictionaries and encyclopedias (although attribution is often required), and the public domain (e.g., the patent corpus). In some cases, extracted content may be automatically paraphrased to avoid copyrights. Like claim support, this kind of mechanical writing will fade away from the task lists of practitioners, who will serve more as editors of this content.

Obtaining Canned Text

Utilizing canned text typically involves a practitioner manually searching through prior work product and copying parts into the document being prepared. Even though there is no fresh writing occurring, there can still be a significant labor cost. One existing system, however, takes an assignee name and a law firm name as inputs and extracts, from the published patent corpus, all the original templates used by the law firm in preparing patent applications for the assignee.[18] This potentially provides a great head start when preparing documents related to past work product. Future systems may add further automation to canned text utilization. For example, the automated extraction of boilerplate and other reused content may be further granularized. Some systems may automate suggestion and/or selection of appropriate reused language for a given project. Again, with canned text, patent practitioners will play a diminishing role and will act more as editors rather than content miners and arrangers.

Document Assembly

Systems exist today that partially automate assembly for patent documents. For example, some systems automatically populate a user-defined application template with auto-generated claim support, letting practitioners skip some of the minutiae of patent drafting. A future system may automatically build application templates or even first drafts based on practitioner input and preferences. Language synchronization may be implemented to ensure that, when the pieces are assembled in a single, well-formatted document, the language throughout is self-consistent. As a prosecution example, existing office action shell generators automatically populate templates with bibliographic information, current claims, standing rejections, etc., letting practitioners get straight to the amendments and/or arguments. Eventually, document assembly should be wrested completely from patent practitioners and delegated to nonpractitioner “document technicians” and/or to automation.

Conclusions

In sum, market forces are requiring patent practitioners to move away from the traditional single resource (i.e., a practitioner), purely manual document production. A content type framework is useful for envisioning patent practitioners’ evolving role in drafting as workflows become more granularized and patent document automation becomes more ubiquitous. In the foreseeable future, practitioners will remain the primary drivers of value creation as they craft bespoke writing content, perhaps with technology acceleration. For the remaining content generation, practitioners will serve mostly as editors of auto-generated and auto-assembled patent content. Individual practitioners will be spared the low-value parts of drafting and will be capable of processing significantly more patent work at the same or better quality as today. Even with a smaller role per document and decreasing fees per project, if patent procurement follows the trend of other automation-disrupted professional services, demand for skilled practitioners will likely increase along with improved access to services, quality, and efficiency.


[1]. Ian Schick, 10 Ways Tech Is Disrupting Patent Procurement, Law360 (May 17, 2019), https://www.law360.com/articles/1159335/10-ways-tech-is-disrupting-patent-procurement.

[2]. Five Favorite Features of LexisNexis PatentOptimizer®, LexisNexis (Sept. 13, 2019), https://www.lexisnexisip.com/knowledge-center/five-favorite-features-of-lexisnexis-patentoptimizer.

[3]. TurboPatent: Helping Companies Protect Their Inventions without Breaking the Bank, Disruptor Daily (May 3, 2018), https://www.disruptordaily.com/turbopatent-helping-companies-protect-their-inventions-without-breaking-the-bank.

[4]. Richard Tromans, Meet Specifio the AI Start-Up Automating Patent Drafting, Artificial Law. (July 28, 2017), https://www.artificiallawyer.com/2017/07/28/meet-specifio-the-ai-start-up-automating-patent-drafting; see also David Hricik, Machine Aided Patent Drafting: A Second Look, Patently-O (Aug. 25, 2017), https://patentlyo.com/hricik/2017/08/machine-patent-drafting.html.

[5]. See Malathi Naya, AI Speeds Patent Process, but Robot Attorneys Still a Ways Off, Bloomberg L. (Dec. 31, 2018), https://news.bloomberglaw.com/ip-law/ai-speeds-patent-process-but-robot-attorneys-still-a-ways-off; see also David Hricik, Augmented Patent Drafting and Ethics, Patently-O (June 8, 2017), https://patentlyo.com/hricik/2017/06/augmented-patent-drafting.html.

[6]. Ryan Abbott, The Reasonable Robot: Artificial Intelligence and the Law 22 (2020).

[7]. Cem Dilmegani, When Will Singularity Happen? 995 Experts’ Opinions on AGI, AIMultiple (Oct. 8, 2021), https://research.aimultiple.com/artificial-general-intelligence-singularity-timing.

[8]. See Jacob Goldstein, How the Electronic Spreadsheet Revolutionized Business, NPR (Feb. 27, 2015), https://www.npr.org/2015/02/27/389585340/how-the-electronic-spreadsheet-revolutionized-business; see also Lisa Cumming, After VisiCalc Revolutionized Accounting in the 70s, AI Is the Next Big Breakthrough, Blue J (Dec. 5, 2017), https://www.bluej.com/ca/blog/single-post/2017/12/05/after-visicalc-revolutionized-accounting-in-the-70s-ai-is-the-next-big-breakthrough.

[9]. See James A. De Lapp et al., Impacts of CAD on Design Realization, 11 Eng’g, Constr. & Architectural Mgmt. 284 (2004), https://doi.org/10.1108/09699980410547630.

[10]. The AIPLA/AIPPI/FICPI AI Colloquium Primer 3 (2019).

[11]. See Ian C. Schick, US Patent Filings: Peaking or False Peak, Faster Pats. Blog (Mar. 6, 2019), https://blog.specif.io/2019/03/06/us-patent-filings-peaking-or-false-peak; see also Ian C. Schick, Patent Practice 3.0, Presentation at the AIPLA Mid-Winter Institute 2020 (Jan. 30, 2020), https://blog.specif.io/2020/01/30/aipla-mid-winter-institute-2020-presentation.

[12]. Ian Schick, What a Maturing Patent Bar Means for the Industry, Law360 (July 9, 2019), https://www.law360.com/articles/1176373/what-a-maturing-patent-bar-means-for-the-industry.

[13]. Am. Intell. Prop. L. Ass’n, 2009 Report of the Economic Survey (2009), https://www.aipla.org/detail/journal-issue/2009-report-of-the-economic-survey; Am. Intell. Prop. L. Ass’n, 2019 Report of the Economic Survey (2019), https://www.aipla.org/detail/journal-issue/2019-report-of-the-economic-survey.

[14]. Ian C. Schick, A Production View on Patent Procurement, IP Theory, Winter 2020, https://www.repository.law.indiana.edu/ipt/vol9/iss1/3; see also Ian C. Schick, A New Paradigm for IP Practice 3.0, Faster Pats. Blog (Jan. 6, 2020), https://blog.specif.io/2020/01/06/a-new-paradigm-for-ip-practice-3-0.

[15]. Ian C. Schick, Understanding Patent Document Automation, Faster Pats. Blog (Oct. 8, 2019), https://blog.specif.io/2019/10/08/understanding-patent-document-automation.

[16]. See Ed Sohn, Biglaw Counsel with Biglaw Needs Develops the alt.legal Solution, Above the L. (Oct. 18, 2017), https://abovethelaw.com/2017/10/biglaw-counsel-with-biglaw-needs-develops-the-alt-legal-solution; see also David Hricik et al., Ethics of Using Artificial Intelligence to Augment Drafting Legal Documents, 4 Tex. A&M J. Prop. L. 465 (2018), https://scholarship.law.tamu.edu/cgi/viewcontent.cgi?article=1080&context=journal-of-property-law.

[17]. See Sys. & Methods for Providing Adaptive Surface Texture in Auto-Drafted Pat. Documents, U.S. Patent No. 11,023,662 (filed Apr. 3, 2020), https://patents.google.com/patent/US11023662B2.

[18]. See Sys. & Methods for Extracting Pat. Document Templates from a Pat. Corpus, U.S. Patent Application No. 16/901,677, Publication No. 20200311351 (published Oct. 1, 2020), https://patents.google.com/patent/US20200311351A1.


Share this article

AIPLA Mid-Winter Institute 2020 (Presentation)

Share this article

Specifio’s Co-Founder & CEO, Ian C. Schick, presented in the main programming of this year’s AIPLA Mid-Winter Institute in a session entitled “Will a Machine Replace Me? The Impact of AI and New and Emerging Technologies on IP Practice”. Below is a link to the slide presentation.


Share this article

A New Paradigm for IP Practice 3.0

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

Image result for new paradigm

A recent blog article by European patent powerhouse, Kilburn & Strode, described a state of transition in the patent industry in the context of three eras: IP Practice 1.0, IP Practice 2.0, and IP Practice 3.0. IP Practice 1.0 was largely paper-based with the predominant technologies including word processing and facsimile. The hallmark of IP Practice 2.0 has been the digitization of patent files and prior art documents, which has enabled electronic filing and searching, as well as newer technologies like machine translation. Now, according to K&S, “we are embarking on the era of IP Practice 3.0, which offers greater automation, more use of AI applications …, and the customisation of other technologies” to “increase efficiency, simplify tasks and offer greater value to clients.” 

We are seeing more and more manifestations of the transition to this new epoch, from the emergence of new business models for patent practices to the explosion of new technologies specifically designed to augment and amplify practitioners and other patent professionals. This article begins with a look at some of the immediate challenges facing IP practices, particularly in patent preparation and prosecution, in the new era as efficiency transformation—through leveraging technology and restructuring operations—becomes mainstream and mandatory to remain competitive. Next, a new framework is proposed for contemplating and structuring patent practices in a way that specifically addresses many of today’s challenges.

Immediate Challenges for IP Practice 3.0

Modern patent practices face several pressing challenges which are in large part the impetus for evolving beyond IP Practice 2.0. The first challenge is simply the economics of practice. Take patent preparation, for example. Despite more demand than ever on practitioners, intensifying fee pressure combined with a drive to fixed fee and other hourly-billing alternatives has resulted in a decade-plus industry trend of decreasing fees for preparing patent applications. 

Complicating the economics even further is a pronounced inversion of available leverage among the patent bar, i.e., the associate model has largely collapsed in present day patent practices. Indeed, the last 10 years have seen a drastic falloff of early-career practitioners. The histogram in Figure 1 below bins active practitioners based on their years of practice. 

Figure 1. Currently active USPTO registered practitioners binned based on years of practice.

Readily apparent in the histogram above is the glut of mid- to late-career practitioners with 10-20 years of practice and the surprisingly low relative numbers of practitioners with less than 10 years experience. As the average practitioner is becoming more senior, it logically follows that the industry average billing rate is also rising—amplifying the effects of declining fees for patent preparation. Fee pressure means time pressure for practitioners; and time pressure means a focus on efficiency and quality assurance is more important than ever.

The growing essentiality of optimizing practice efficiency without compromising quality has brought about a recent boom in available “PatentTech” offerings for practitioners and other patent professionals. That in itself, however, has created other challenges for transitioning practices, specifically around procurement and adoption of these new technologies. For example, today’s modern tool offerings for patent practices are highly fragmented often with extremely narrow solutions. This means discovering and vetting available tech can be a cumbersome chore and achieving comprehensive, or even broad coverage across the patent pipeline may likely require engaging a dozen or more vendors. Too many vendor offerings also makes garnering stakeholder buy-in more difficult, particularly at law firms with many partners. 

A Proposed Practice Framework for IP Practice 3.0

Traditionally, legal practice structures are chiefly based on seniority, e.g., partners, associates, agents, paralegals, secretaries, and so on. As technology and efficiency play increasing roles in addressing the challenges discussed above, however, the focus for defining patent practice structure needs to shift to a more modern approach centered on business function. Figure 2 provides a visualization of a proposed structure for 3.0 patent practices—a trifurcated framework comprising a Services Layer, a Production Layer, and a Counseling Layer as separate but interdependent business units. The practice functions were grouped in this way specifically because the activities associated with each layer are conducive to very different approaches for optimization.

Figure 2. A proposed framework for patent practices in the era of IP Practice 3.0.

Several leading patent practices are already incorporating versions of this philosophy in their organizations. The different layers have distinctly different functions, requiring different skill sets and different personnel. While many patent professional roles will change to varying degrees, the proposed framework provides a more efficient alignment of human resources with practice aims and goals.

      Services Layer

In the new era of IP practice, everything that can be automated should be automated. That includes all activities traditionally performed by humans which are mechanical and repeatable and do not involve significant creativity or judgment. In addition to automation, insights from data analytics should be leveraged for most, if not all decisions. For the purposes of this discussion, these automatable or data-driven tasks and workflows are described as “services” in IP Practice 3.0 model. The associated technologies form the bedrock of the practice structure as the Services Layer. 

While the Services Layer is primarily tech-based, some personnel are needed for procurement, implementation, and ongoing operation. To address the issue of adoption, practices may be wise to appoint specific practitioners or other individuals to train on and operate the various technologies in the Services Layer instead of pushing loads of new technology on the entire practice.

Automatable activities can be grouped based on whether they are traditionally performed by registered practitioners or by non-practitioners, such patent secretaries, paralegals, docketers, and other patent professionals. Many non-practitioner activities are excellent candidates for automation including meeting scheduling, preparing filing forms, filing documents with the Patent Office, client reporting, docketing, invoicing, information disclosure statement (IDS) management, shell generation, proofreading, and more.

While the availability of code-free robotic process automation (RPA) tools makes many automations easy enough for non-technical individuals to set up, some technical skills are likely needed to string together all the various systems and workflows of a busy practice. As for patent professionals whose traditional roles are displaced by automation, they will still be in demand. For some, their knowledge and expertise will be indispensable in assisting with the design and implementation of patent practice automation systems. Many, however, will find new roles within IP Practice 3.0, primarily in the Production Layer, described further below.

When it comes to activities traditionally performed by registered practitioners, technology in IP Practice 3.0 plays a supportive role, rather than a supplantive one. Data analytics support practitioners as they perform analysis and formulate advice. Automation comes into play during the preparation of substantive patent documents including patent applications, office action responses, briefs, etc. In sum, technology should augment practitioners’ tasks with intelligence and should remove as much rote and mundane work from practitioners’ plates as possible.

      Production Layer

The Production Layer is responsible for generating high-volume substantive patent documents, under the direction and oversight of the Counseling Layer. The Production Layer leverages supportive technologies from the Services Layer to help a diverse team of patent professionals generate most of the practices’ work product. Segregating the production function allows IP Practice 3.0 to fully realize the benefits of lean production principles. According to lean production thinking, processes should be viewed as a series of separate but interlinked subprocesses. Each subprocess is then delegated to the most efficient resource. 

In IP Practice 3.0, registered practitioners operating in the Production Layer should be used to draft substantive input for the documents being generated, but that is about it. Most other subprocesses in document assembly can be automated or delegated to other patent technicians (e.g., former patent paralegals and secretaries). This reduces overprocessing—using an expensive resource for a task when another resource could complete the task just as well—a key tenet of lean production. 

To illustrate a lean process, consider patent application preparation. A registered practitioner may be used to draft only certain components of the application like the background section, the problem/solution statement, a primary claim set, and descriptions of drawings illustrating example invention embodiments. Based on the primary claim set as input, natural language generation (NLG) can be used to create content for substantial portions of the patent application. Taking what the practitioner drafted along with the auto-generated content, a patent technician can then assemble a close-to-final draft of the patent application for review and finalization by the Counseling Layer.

Like any production environment, work product consistency in IP practices is crucial for maintaining customer/client expectations and maximizing production efficiency. Strict process and document parameters set by the practice (and not by individual practitioners) can ensure consistency while avoiding overproduction—generating more work product than is needed to meet a client’s needs—another key tenet of lean production. Certain supportive technologies, like automated proofreading and checklists, can minimize defects and streamline quality assurance.

​​      Counseling Layer

The specific functions associated with the Counseling Layer are functions that more or less remain the same in IP Practice 3.0 as they were in traditional law firm and in-house practices. They include most common practitioner activities—except for generating high-volume substantive patent documents (which is handled by the Production Layer). Accordingly, the Counseling Layer is comprised mostly of practitioners but also some number of administrative and support roles.

An obvious first category of functions covered by the Counseling Layer broadly includes activities requiring direct client interactions. Some examples include client relations, counseling sessions, invention disclosure meetings, conveying advice and work product, etc. Here, practitioners are the “face” of the practice.

A second functional category handled by the Counseling Layer is high-end legal services like portfolio strategy and management, patentability and infringement analyses and opinions, IP due diligence, and so on. These things are often too low-volume and too bespoke to gain much leverage through automation or delegation. That is, high-end legal services are high-end precisely because they involve original analysis on novel fact sets and cannot be mass produced. 

Finally, the Counseling Layer is responsible for directing and overseeing work product generation by the Production Layer. This includes providing document instructions to the Production Layer as well as performing pre-filing review and finalization of all substantive patent documents generated by the Production Layer. 

Conclusions 

IP Practice 3.0 is the new reality. Shifting practices to a functional organizational structure aligns human resources such that talent is specifically focused only where it is needed. Leverage can then be efficiently maximized through automation and delegation. This helps address the surplus of mid- to late-career practitioners and the dearth of early-career practitioners while coping with technology-driven attrition of other patent professionals. 

By separating document production from counseling and basic services, lean production can be used to optimize the value of a practice’s work product. The result is prices and quality that satisfy the market without destroying practice cost models. In addition, limiting the use of new technologies to a select group of professionals, instead of the entire practice, can help address issues around adoption.

Finally, new business models are already emerging in the era of IP Practice 3.0 focusing on specific layers or parts of layers in the proposed practice framework. These new business models include examples in law firms and in-house patent departments. They also include companies selling into the patent industry with outsourced technology and talent. The benefits of economies of scale are obvious for services but also for production. By outsourcing some or all of the Services and Production Layers, it’s foreseeable that relatively small practices serving as the Counseling Layer will be able to move huge amounts of patent work.

DISCLAIMER: The views and opinions expressed in this article are those of the author and (1) are not provided in the course of and do not create or constitute an attorney-client relationship, (2) are not intended as a solicitation, (3) are not intended to convey or constitute legal advice, and (4) are not a substitute for obtaining legal advice from a qualified attorney. You should not act upon any such information without first seeking qualified professional counsel on your specific matter. The hiring of an attorney is an important decision that should not be based solely upon Web site communications or advertisements.



Share this article

Understanding Patent Document Automation

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

For attorneys, gaining a practical understanding of technologies impacting the practice of law is no longer just an option. The rules of professional conduct in most US states now explicitly require a duty of technology competence, primarily so that attorneys can (1) determine whether and when to incorporate technologies into their own practices and (2) adequately counsel their clients on the benefits and risks associated with these technologies. In states that do not have the explicit requirement, the duty of technology competence is generally considered to fit implicitly within the existing duty of competence requirements. To be sure, the duty exists always, not just if an attorney decides to leverage technology in their practice.

In patent practice, the industry is experiencing an explosion in new technologies that let practitioners process their work efficiently with less errors, make data-driven decisions, and, overall, provide more value to their clients while respecting existing budget constraints. Indeed, tech adoption is becoming a competitive imperative for patent practices in order to cope with challenging economic and demographic trends in the hyper-competitive patent industry.

At the recent Intellectual Property Owners Association (IPO) Annual Meeting in DC, the general session that opened the conference was titled “IP Automation – What’s Here Today, Not Years Away?” Representatives from private practice and in-house presented ways they are automating processes (e.g., docketing, filing, etc.) and automating patent preparation. The latter made a huge splash and dominated the buzz for the rest of the conference.

This article seeks to provide a practical understanding of document automation in legal, generally, as well as in patents, specifically.

What is Document Automation?

Document automation (also known as document assembly) relates to technologies designed to assist in creating electronic documents. Typically with some amount of human input, computers assemble text and other content into new documents. The text used to build the documents may be pre-existing (or “canned” such as boilerplate or templated text) or it may be computer-generated on the fly during the assembly process.  

Document automation has been on the rise for years across many industries and is now becoming mainstream in legal. Stanford’s LegalTech Index lists about 250 legal document automation companies with more and more being added regularly. These companies serve virtually all areas of legal practice.

Benefits and Risks

The benefits of contemporary document automation generally far outweigh the potential risks. For example, document automation reduces labor needs for rote and mundane writing. Not only are these types of tasks detrimental to attorney job satisfaction, it is also wasteful to have a highly-trained practitioner performing tasks like this at steep hourly rates. Reducing burnout saves in recruiting and training costs. And reducing waste is essential for patent practices competing in today’s market.

Document automation has benefits beyond labor savings in document drafting. It reduces risks associated with human error. Because of this, and the ability to create more consistency between documents of the same type, document automation can drastically reduce the amount of time needed for proofreading. Since proofreading often rests with law firm partners or time-strapped in-house attorneys, time savings here is particularly impactful.

Some of the risks associated with document automation are the same as with any technology for attorneys, such as maintaining confidentiality and data security. Here, vendor transparency is critical so that attorneys can accurately assess these risks and counsel their clients accordingly. 

Other risks relate to adoption and utilization. For example, failure to properly use or leverage technology where the costs are passed on to the client can potentially result in overbilling. If the cost being passed to the client does not reflect the full value that should be realized by using the technology, then the fees may be considered unreasonable from an ethical standpoint.

Another risk might be improper reliance on a technology due to a misunderstanding of its capabilities. For example, if an attorney relies on a technology to perform tasks A and B, but the technology is only designed to perform task A, then task B is either not being done at all or it’s being done inadequately, and potentially without the attorney realizing it.

A Framework for Understanding Document Automation

To have a practical understanding of how and to what extent a given type of legal document is or could be automated, it is useful to decompose the document based on content type. Legal documents generally have some combination of bespoke writing, mechanical writing, and canned text.

Bespoke Writing

Bespoke writing reflects the intellectual heavy lifting performed by attorneys while preparing legal documents. It often involves original analysis or disclosure on a unique fact set. This type of writing requires creativity and judgment. It is typically guided by the attorneys’ training, experience, and their knowledge of their client’s business and business strategy. Bespoke writing is too nuanced and context-dependent to be automated with existing technologies. This is where attorneys provide their primary value-add to written work product, which will likely remain the case for the foreseeable future.

The world’s most advanced system for natural language processing (NLP) and natural language generation (NLG), by far, exists between your ears. Expecting a machine to do bespoke legal writing is unrealistic with today’s technologies. Some cutting-edge technologies can generate text that may appear bespoke, but it can be unpredictable and nonsensical in the context of a specific legal document and underlying fact set. Recurrent neural networks (RNNs), for example, can create never-before-seen text, but only in relatively short spans (under 100 words). RNN plus neural bag-of-words (BoW) is used for real-time predictive text (e.g., Gmail’s Smart Compose), but is currently limited to just a few words ahead of a typist’s cursor.

Mechanical Writing

Mechanical writing is the traditionally rote and mundane parts of legal writing projects. This writing is usually driven by convention and/or by satisfying document requirements. It must be accurate and complete, but does not require significant mental work. Since the mechanics of mechanical writing are essentially repetitive across different documents of the same kind, this type of content in legal documents is ripe for automation.

From a technical perspective, mechanical writing can be automated with some combination of text transduction, text extraction, and text generation. Text transduction simply means turning one span of text into a different span of text. A common example of this in legal writing is propagating certain language throughout a document. If it were done manually, an attorney might copy and paste text into different parts of the document and then massage it so that the text reads appropriately for the different parts.

Text extraction relates to locating and describing facts to provide context and support within a document. In legal writing, extracted text may come from resources such as dictionaries, rules and statutes, encyclopedias, and other written bodies of knowledge. When used for document automation, relevant text (e.g., sentences or paragraphs) may be lifted from existing documents and used as content in newly generated documents.

Mechanical writing can include text generation such as summarization and data-to-text. With summarization, the most salient points of a larger document are distilled into a brief passage, which can be automated. One technique for automatically identifying pertinent terms in a document is known as “term frequency–inverse document frequency” (TF-IDF). This approach involves measuring, for each term used in a given document, the frequency in which the term is used in written language, generally, and then dividing that by the frequency of the term in the document itself. If a term is normally used rarely but appears often in a given document, the term is likely an important one.

When it comes to data-to-text, classic examples include weather reports, sports reports, and stock reports where noteworthy aspects of numerical data is automatically expressed with language. Because legal writing is not often informed by numerical data, data-to-text is less common in legal document automation.

Canned Text

Canned text is predetermined language such as boilerplate and template text. While this type of text is pre-existing, there is still an associated labor cost with manually identifying and locating appropriate canned text (e.g., in a template repository, old related documents, etc.). Different canned text needs to be properly organized within a document. It also may need to be adapted for specific projects, for example, by completing variable or conditional parts of the canned text.

Luckily, computers do a very good job of dealing with canned text for the purposes of document automation. A simple example is mail merge for letters and simple documents. Here, given some basic input information, a document with templated text is automatically completed by essentially filling in the blanks. A more complex example is contract assembly. With automated contract assembly, an attorney may identify a type of contract as well as the basic facts and terms. Based on that information, standard-language clauses are assembled into a draft contract document.

Different Legal Documents Lend Themselves to Different Automation Technologies, or Not at All

Depending on the type of legal document, there will be different ratios of bespoke writing, mechanical writing, and canned text. The ratio for a given type of legal document determines whether or not the document is a good candidate for automation and, if so, which technologies are best suited to automate that type of document.

At one end of the spectrum, contracts are dominated by canned text. There may be some amount of mechanical writing (e.g., propagating certain facts throughout the document) or bespoke writing (e.g, novel terms), but, by and large, contracts are mostly standard language. On the other end of the spectrum, memoranda are often primarily composed of bespoke writing where original analysis is being performed on a unique fact set. As such, they are generally not candidates for automation. Most legal documents, however, will fit somewhere in between where mechanical writing makes up a significant portion.

Document Automation in the Patent Context

A combination of statutory requirements and case-law-informed drafting conventions drive the contents and structure of patent applications. To be sure, a significant component of patent preparation includes bespoke writing where key concepts and nuances of an invention are described in a careful and strategic manner. The claims, obviously, fit squarely into this category, as do the background section and problem / solution statement.

A substantial component of most patent applications, however, includes mechanical writing and canned text. These are the parts that can be readily automated with currently available tools. In a patent application, mechanical writing may cover things like the title, field of the invention, summary, literal claims support in the detailed description, additional claim sets mirroring attorney-written claims, abstract, and basic drawing figures like those illustrating the environment in which the invention is practiced and method flow charts. 

Other examples of mechanical writing may include lists of well-known examples (e.g., “Examples of types of material strength may include compressive strength, shear strength, tensile strength, and/or other types of strength.”), dictionary definitions (e.g., “Tensile strength is the resistance of a material to breaking under tension.”), and descriptions of well-known facts (e.g., “Steel is an alloy of iron and carbon, and sometimes other elements.”). 

Canned text in patent applications often includes things like boilerplate, stock figure descriptions, and stock definitions.

Conclusions

The duty of technology competence is now widely adopted in the US and applies regardless of whether tech is actually adopted. The threshold is not expertise, but rather knowing your limits and asking for help when appropriate so that you can make determinations about adopting tech in your practice and counseling clients as to the benefits and risks.

Document automation is becoming increasingly relevant to legal practice, particularly with patents. It helps practitioners avoid rote and mundane writing, which benefits them and their clients alike. With more and more products coming to market, it is important to have a practical understanding of document automation to have realistic expectations and appreciate the potential for your practice.

DISCLAIMER: The views and opinions expressed in this article are those of the author and (1) are not provided in the course of and do not create or constitute an attorney-client relationship, (2) are not intended as a solicitation, (3) are not intended to convey or constitute legal advice, and (4) are not a substitute for obtaining legal advice from a qualified attorney. You should not act upon any such information without first seeking qualified professional counsel on your specific matter. The hiring of an attorney is an important decision that should not be based solely upon Web site communications or advertisements.


Share this article

Patent Work Product: A Reflection Of Your Firm’s Brand

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

Should your firm’s patent work product be a part of the firm’s brand strategy? How does having a “house brand” for work product matter for law firms involved with patent procurement? These are some of the topics covered in this article but, before getting into the weeds, let’s first take a step back and discuss branding for patent law firms more generally.

What is “brand” when it comes to patent law firms?

In the broadest sense of the word, your firm’s brand could be thought of as a collective mental construct associated with your firm. It is the overall impression of the firm felt by clients, prospects, and the public at large. It includes everything that comes to mind when one thinks of your firm—both factual (e.g., firm name, logo, attorney pedigree, types of clients, etc.) and emotional (e.g., reputation, quality, value, prestige, etc.). 

Why is brand consistency important for law firms?

Brand consistency refers to a pattern of expression affecting perception about your firm. The more consistency in the pattern, the more consistent the brand, which is important on a variety of fronts. For example, a cohesive law firm brand projects professionalism as compared to a brand expression that is all over the place. With professionalism comes trust and loyalty. Brand consistency encourages confidence among law firm clients that they’ll dependably receive a certain level of service when they engage your firm.

A well-defined brand image acts as a guide for marketing and branding decisions. It is crucial for shaping your firm’s brand perception. Only when all brand elements are coordinated and complementary can a firm’s brand be reliably shaped in the minds of clients and others. And it’s not just important for people outside of your firm. A consistent brand facilitates internal direction by aligning attorneys and staff with the firm’s values and positioning.

One of the most important functions of brand consistency is for differentiating your firm from other patent law firms. In competitive markets with near-identical offerings, a clear and consistent brand is what sets your firm apart while attracting new clients and helps retain existing ones.

Is patent work product an element of a law firm’s brand?

Your firm’s brand is what it projects to the world outside your firm, most crucially to the firm’s existing clients. Along with direct client communications and billing, the provisioning of work product is one of the most important interactions between a firm and its clients. Substantive work product for patent law firms primarily includes patent applications, office action responses, and briefs.

The work product itself is the culmination and results of what your firm’s brand stands for. It represents delivery on the promise of a level of service your clients pay for and expect. As such, to support brand consistency, the work product coming out of your firm should consistently reflect your firm’s brand–just like any other brand element.

Who controls brand–the patent law firm or individual patent attorneys?

Imagine if every cook at McDonald’s had their own twist on the Big Mac? Customers would never know what they were going to get unless they only went to McDonald’s when their “favorite” cook was there. But even with their favorite cook, they may never be content and may instead be constantly wondering whether they could be getting a better sandwich from a different cook at the same restaurant. While this analogy may be silly, it is exactly the type of situation many patent law firm clients find themselves. 

A law firm client cannot be consistently assured they are getting the value they’re paying for if the work product is not well-defined. Because one client’s patent cases are often handled by a team of practitioners with ranging experience levels and backgrounds, the work product they receive can vary widely depending on who was responsible for preparing it. That is, unless the firm’s work product is managed to provide a consistent work product each and every time.

What about the argument that diverse work product is actually an advantage when it comes to patents because it acts as a hedge against future changes in laws affecting patent interpretation and validity? To some, this may seem more like an unprovable client pacifier than savvy business advice. While sounding logical and soothing on its face, the argument reflects a dangerous underlying position that a firm’s attorneys should not do what the firm regards as best practices and should instead “do their own thing” when it comes to work product.

Because the business value of brand consistency in work product likely outweighs any potential benefits of diverse work product, patent law firms may be wise to consider dictating firm branding as it relates to work product, instead of leaving it in the hands of individual attorneys.

How to make patent work product consistent across practitioners such that it supports brand consistency?

A variety of strategies can be deployed to promote consistent work product across a firm. First and foremost, however, it is essential to centralize the management of work product branding in a way that is accessible to individual attorneys. The firms should establish what it regards as best practices to guide practitioners as they generate work product. 

Templatization

Among chefs, there’s a saying that diners eat first with their eyes. The same goes for law firm clients when they’re evaluating your work product. The appearance of the documents is a reflection of your firm’s brand. A clean-looking and well-organized document will always be received better than a sloppy or inconsistent one. Templates are a powerful tool for ensuring a consistent look and feel to your firm’s work product. Often, even template formatting can be augmented and changed by an individual attorney’s copy and paste actions, resulting in an ugly work product that is scrutinized more heavily based on the appearance of the document, let alone the content. Thus, to work effectively, templates need to be maintained by a single person or group of people. The benefits of templates also extend to efficiency and quality control. For example, with centralized template management, firmwide updates to boilerplate can be effectuated instantly and consistently for all practitioners.

Standardized document parameters

Having well-defined and enforceable document parameters is another key element for generating consistent work product. Take a patent application, for example. Firm mandated document parameters may include things like the time spent taking in and studying each invention prior to drafting the corresponding application, the number of claims, the types of claims, the time spent drafting claims, the number of figures, the types of figures, the time spent preparing figures, the length of specification, the sections to be included in the specification, the structure and language used for each individual section in the specification, time spent preparing the specification, and so on.

Document parameters can extend to document-related processes as well. For example, there may be prescribed client approval checkpoints such as after the claims are prepared but before work has begun on the specification and figures.

Successful implementation of a well-defined set of document parameters will likely depend on a project intake process that includes categorizing new projects based on client, technology, and/or complexity. This is necessary to align each project with the appropriate parameters. An organized intake process may also complement the alternative fee arrangements (e.g., tiered flat fees) becoming ever more popular.

Why do law firm clients value consistent work product?

From an in-house practitioner’s perspective, consistent work product means knowing ahead of time what they’re going to get in return for each work assignment to outside counsel. That, in turn, means less review time and peace of mind knowing that they won’t be hung out to dry.

As for the patent-owner company, the value of consistent patent work product may be more along the lines of transparency into the strength of their portfolio. If a portfolio manager is confident in what is consistently portrayed in the underlying patents, then there is more clarity when it comes to opportunities for leveraging the portfolio. Indeed, work product consistency is so important for some sophisticated consumers of patent services that they prescribe their own application templates and stringent document requirements to their outside counsel.

How do patent law firms benefit from delivering consistent work product?

The benefits of consistent work product are not limited to brand perception and the value felt by firm clients. The patent law firm itself also benefits. For starters, it encourages systematization of document preparation, which in turn encourages efficiency. Despite the billable hour, law firm efficiency allows for “doing more with less” and coping with challenging economic trends in the patent market. 

Consistent work product lets quality control be streamlined and tightened. For example, if the underlying document template is already firm-approved, the reviewer need only review the substantive parts of the document that differ from similar prior documents. Well-defined document parameters lend predictability to the work product itself, not to mention delivery times and the law firm’s cost associated with generating the document.

Conclusion

It is not controversial that brand consistency is one of the central tenants to modern businesses. Why should patent law firms differ? Law firm brand encompasses work product and, therefore, consistent work product across a firm is an essential element of successful branding. The stronger the brand, the easier it is to win new clients and keep your existing ones happy and loyal to the firm. 

DISCLAIMER: The views and opinions expressed in this article are those of the author and (1) are not provided in the course of and do not create or constitute an attorney-client relationship, (2) are not intended as a solicitation, (3) are not intended to convey or constitute legal advice, and (4) are not a substitute for obtaining legal advice from a qualified attorney. You should not act upon any such information without first seeking qualified professional counsel on your specific matter. The hiring of an attorney is an important decision that should not be based solely upon Web site communications or advertisements.


Share this article

A Production View on Patent Procurement

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

When we think of a “production environment”, a law firm patent practice is not usually the first thing that comes to mind. But why not?! Patent practices are highly process-oriented and they certainly involve “manufacturing” work product, primarily in the form of new patent applications and office action responses. This article discusses how, with a production view on patent procurement, exploiting the principles of lean production can be a compelling way to adapt to tough issues presently roiling the patent ecosystem.

The idea of commercial manufacturers providing completely-handcrafted products went out the window, for the most part, during the industrial revolution (the first one). To be sure, examples of prideful handiwork can still be found today, like with Amish woodworking or specialty items like, say, Rolls Royce automobiles. Speaking of cars, Rolls Royce and BMW both had record sales last year. Rolls Royce automobiles are assembled by hand and are of the finest quality. BMW automobiles are also very high quality, but manufactured by employees, automation, and OEMs working in concert. Rolls Royce sold 4,107 vehicles in 2018 (0.082 vehicles per employee), while BMW sold over 2.49 million vehicles (18.5 vehicles per employee). If these car companies were law practices and the cars were patent applications, which one would you want to model your own practice after?

We all know the Rolls-Royce approach to patent procurement–more or less the way it’s been done for the past 100 years. In today’s hyper-competitive patent market, however, having a more BMW-style practice may prove key to prospering (or even to survival in some cases). So what would a BMW-style patent practice look like? More specifically, how can patent practices modernize through lessons learned in traditional production industries–where decades of intense competition has resulted in process optimization evolving into a science?

Some of the most successful patent firms today are leveraging ideas from “lean production” to maximize their competitive edge. Lean production is a convenient framework for thinking about operations and possible improvements. It is a school of thought that originated in the Japanese automotive industry in the 1990’s. The basic idea is to maximize the creation of value for customers while eliminating waste. Here, “value for customers” means any action or process that a customer will be willing to pay for. Parts of the production process that do not add value are considered waste.

Several interrelated concepts are central to the lean production ethos. They include minimizing waste, just-in-time production, kaizen (continuous improvement), and cell production. Each of these is explained below in the context of patent procurement.  

Minimizing waste

Waste in patent procurement can take on many forms. At a high-level, however, waste can be categorized as overprocessing, overproduction, or defects.

Overprocessing

There is a waste of resources if an expensive resource (e.g., an attorney, a patent agent, a paralegal, a secretary, etc.) is used for a task when another resource could complete the task just as well. At first blush, this sounds a lot like the classic mantra of law firm leverage. However, there are things that can be done besides delegating talks to lesser-trained individuals to avoid overprocessing. For example, tasks can be decomposed into subtasks and those subtasks can be examined for further delegation even if the larger task is traditionally handled entirely by the more expensive resource. Also, in some cases, emerging technologies in automation may present alternatives to delegation. Where automation can be utilized instead of human labor, tasks are often performed cheaper, faster, and with less errors. 

Overproduction

In the law practice context, overproduction means generating more work product than is needed to meet a client’s needs. Take patent application preparation, for example. How does a drafter know when a patent application is done? In traditional practices, it’s often done when the budget is exhausted. What if, instead, the drafter stopped working on the application when it (1) provides sufficient backup positions that might actually end up in the claims during prosecution, (2) provides sufficient enablement for the initially-claimed and potentially-claimed embodiments, and (3) conforms to any and all requirements from the client. Law firm clients expect work product that satisfies official and strategic requirements, not a treatise on the field of invention.

In order to cut out overproduction, practice management will need to align their own incentives with those of the practitioners doing the drafting work. One easy change, for fixed-fee projects, is to give practitioners a fixed billable hours credit for completing the project, regardless of whether it came in under the budgeted time. Rather than just hours worked, which encourages overproduction, this ties performance more closely to revenue generated.

Defects

This includes any mistakes that occur in the patent pipeline, whether they be clerical (e.g., errors in filing forms), more substantive (e.g., curable § 112 issues), or procedural (e.g., incurring extension of time and other unforced fees). Defects-type waste can lead to additional cost in the form of penalties and legal fees for curing mistakes. Prosecution can also be delayed while defects are addressed, meaning even more waste.

Adjustments in processes and leveraging technology can effectively reduce defects. For example, getting religious about the “four eyes principle” can have a drastic impact in catching errors, but perhaps at the expense of efficiency. Today, there are many automated patent proofreaders available that, in just seconds, can thoroughly review a patent application or office action response for common errors.

Just-in-time production

In general, the focus of just-in-time production is on reducing inventory waste. That is, products are not stockpiled, but rather produced “just in time” to meet orders. With minimal stockholding, producers can be more flexible. For example, they can switch to make new products without having to get rid of much stock, meaning they can act quicker to add value.

Inventory in the law-practice context is work that has been requested by clients (i.e., “work orders”) but not yet completed. In traditional practices, where a stable headcount means production capacity (i.e., full utilization of all employee resources) is essentially fixed, completion of any overage work (i.e., work orders above and beyond what employees can process at a given time) is delayed until work orders dip below the production capacity of the practice. A goal of this conventional approach is to compensate for the ebb and flow of work coming in so that practitioners keep as close as possible to their individual capacities (e.g., 35-40 billable hours per week). The inherent side effect of the traditional model, however, is the stockpiling of inventory in the form of pending work orders.

To realize just-in-time production in a patent practice, flexible production capacity is required. Practices should be able to handle bursts in work orders without having to delay completion for lack of available resources. This is done by incorporating non-employee resources into patent workflows. Non-employee resources can include automation tools, contract patent professionals, and/or domestic or offshore outsourcing services. These are resources that can effectively be turned “on” and “off” as needed such that overage work is completed on pace with client work orders while keeping employee resources at their production capacity. All of the undulations in work orders are absorbed by non-employee resources, thus sufficient resources are always available to meet production needs but the practice is never punished by having to pay for unused or underutilized resources.

For work orders that come today, when is “just in time” to complete them? For a variety of reasons, conventional wisdom says filing by client deadlines (e.g., a product release), non-statutory deadlines (e.g., three-month deadline for office action responses), or statutory deadlines (e.g., on sale bar) should not be the goal. Instead, with a lean production approach, work should be completed as expeditiously as possible to minimize inventory, i.e., the time between the client requesting the work and the work being finished. With the rapidly expanding range of available non-employee resources, be it automation or outsourcing, it has never been easier to find right-fit services for patent procurement production.

Kaizen (continuous improvement)

This is really process optimization but in an incremental fashion and in a way that involves all members of a patent practice, top to bottom. The philosophy of kaizen relies on a continuous effort to improve production. To be effective, a decentralized organizational structure is required with regular team meetings to identify and implement small, often quite simple, improvements to processes and activities. Discrete steps may be eliminated, combined, automated, outsourced, or made more efficient in other ways. Providing training to practice members to help them be more analytical may boost results. 

Another requirement for effective kaizen is that efficiency among practice members must be incentivized. As with overproduction, the traditional billable-hours model for assessing practitioner performance can be counterproductive for kaizen. Here again, tying compensation to revenue generated may often get closest to aligning the interests of management and other practice members. One way to do this, as mentioned previously, is awarding practitioners the full balance of the hours budget for fixed fee projects even when they come in under.

Cell production

According to lean production thinking, processes should be viewed as a series of separate but interlinked subprocesses. Each subprocess is then delegated to the most efficient resource. Traditionally, in manufacturing, each process is assigned to a “cell” or group of production workers. In patent processes, a cell could be comprised of one or more employee resources, one or more non-employee resources (i.e., automation tools, contract patent professionals, and/or domestic or offshore outsourcing services), or combinations thereof.

Processes large and small can be decomposed and optimized in this way. At one end of the spectrum, the “process at issue” could be an overarching process (e.g., “starting with an idea, procure a patent”). In fact, some leading practices have one group of practitioners who draft new patent applications and a completely separate group of practitioners who prosecute pending applications before the USPTO. As the process at issue becomes more granular, however, the opportunity is enhanced to squeeze out more value and eliminate more waste.

To illustrate, if the process at issue is preparing a patent application, subprocesses may include something like (a) invention download, (b) drafting claims, (c) assembling a first draft of the specification with baseline § 112 support, (d) bolstering the draft specification with strategic additions for future prosecution and litigation, (e) preparing formal drawings, and (f) finalizing the application ahead of filing. An attorney can certainly perform all of these subtasks, like in a Rolls-Royce-style practice, but it involves a significant amount of “overprocessing”, in the lean production sense. A more BMW-style practice would find the most efficient solution to each subprocess in order to maximize the creation of value for customers while eliminating waste.

Conclusion

With changing demographics among practitioners, challenging economics in the patent market, and exciting new technologies designed for patent practices, it is imperative that practices evolve their operations to remain competitive. To some, transitioning an operating patent practice to incorporate the tenets of lean production may sound like repairing a car while driving down the highway. But it can be done, it has been done, and an incremental approach will keep the changes relatively painless.

DISCLAIMER: The views and opinions expressed in this article are those of the author and (1) are not provided in the course of and do not create or constitute an attorney-client relationship, (2) are not intended as a solicitation, (3) are not intended to convey or constitute legal advice, and (4) are not a substitute for obtaining legal advice from a qualified attorney. You should not act upon any such information without first seeking qualified professional counsel on your specific matter. The hiring of an attorney is an important decision that should not be based solely upon Web site communications or advertisements.


Share this article

What A Maturing Patent Bar Means For The Industry

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

A seismic shift is occurring in the demographics of the US patent bar that will have a lasting impact on the broader patent ecosystem. Specifically, the average patent practitioner is aging, which, in combination with other market forces, is already precipitating a number of fundamental changes in how patent practices must operate to remain profitable.

Below, I present a histogram of the number of currently active practitioners binned based on their earliest year of registration at the USPTO. For example, if a given practitioner registered first as a patent agent and then again a few years later as a patent attorney, this graph would reflect the year registered as a patent agent. To make this information a little easier to pick apart, we’ve divided the active practitioners into 5-year cohorts.

Before diving into the analysis, I should mention it was performed under the assumption that the underlying data from the USPTO Active Practitioner Roster is at least close to accurate with respect to the active status of practitioners. Also, Cohort #1 represents 4.5 years at the time of writing and will likely grow by another roughly 500 practitioners in the remaining 6 months concluding at the end of 2019. The discussion below, however, considers the numbers as of June 2019.

Not surprisingly, there are major falloffs after 20, 25, and 30 years of practice, which likely coincide with retirement in Cohorts #5, #6, and #7, respectively. The more interesting trend, however, lies with early- and mid-career practitioners.

In a workforce reflecting the typical law firm leverage model, we would expect to see the highest numbers of practitioners in the most recent years of registration. Instead, we see levels that are not even close to the highest. Cohort #2, for example, is 65% larger than Cohort #1. Cohorts #3 and #4 are both over twice the size of Cohort #1. Twice! 

But it doesn’t stop there. Cohort #1 is even smaller than Cohort #5. Stated differently, there are more active practitioners today with 21-25 years of practice experience than there are active practitioners with 1-5 years experience. Indeed, half of all active practitioners have been practicing for 16 or more years, with the average years of practice for all active practitioners at 17 years. This should be shocking for patent owners and patent practice leaders, alike.

“There are more active practitioners today with 21-25 years of practice experience than there are active practitioners with 1-5 years experience.”

All of this begs the question: Why is Cohort #1 so surprisingly small? Is it less STEM students, higher engineering salaries, law school falling out of vogue, bad press from Alice/Mayo, who knows? There are probably many factors at play which certainly deserve a closer look.

In the near term, the demographic shift we’re experiencing as an industry is leading to some very tough issues. For example, average billing rates are rising at an accelerated pace because the average practitioner is becoming more senior in combination with the usual rate increases for inflation.

For patent drafting, higher hourly rates means either lower hours-budgets per app or higher fees per app. The latter seems very unlikely given the ten-year industry trend of stagnant, if not declining, average fees for preparing a patent application. Lower hours-budgets per app is also problematic, in that it can only go so far until the effects on patent quality are unacceptable. Furthermore, to the extent that rising average billing rates are driving reliance on the junior practitioners of Cohort #1 for certain types of work like patent drafting, we can expect to see the shortage of practitioner bandwidth becoming increasingly acute. 

Even more ominous is what lies 5 or 10 years down the road. As mentioned above, retirement trends manifest as falloffs in active practitioner numbers at 20, 25, and 30 years of practice. Assuming average career lengths stay more or less constant, we can expect the retirement trends to hold true with Cohorts #3 and #4. If there is a drastic reduction in numbers in these cohorts, who will be there to replace them? Not Cohorts #1 and #2! 

Now consider the fact that annual filings at the USPTO have been on an ever-increasing trend for the last 30 years. We, as an industry, cannot continue to defy the basic principles of economics forever. This just does not make sense: increasing demand (i.e., annual patent filings), decreasing supply (i.e., practitioner bandwidth), and decreasing costs (i.e., stagnant/decreasing fees).

Let’s assume that market forces will keep patent demand and costs on their current trajectories. That leaves supply, in the form of practitioner bandwidth, as the thing that must change to balance the supply-demand equation. One way to increase practitioner bandwidth would be to simply add significantly more new practitioners to the workforce. That, however, seems unlikely and, even if it did happen, it would do nothing to change the low numbers in Cohorts #1 and #2.

Another option for increasing practitioner bandwidth is to increase practitioner efficiency. With patent drafting, for example, efficiency can be looked at in two ways: dollars per application or hours per application. The dollars-per-app viewpoint is what is driving law firms to leverage outsourcing and insourcing in a quest to find lower-cost ways of generating patent application documents in the traditional manner (i.e., written and reviewed manually).

Patent outsourcing typically means outsourcing drafting work to a domestic alternative legal service provider (ALSP) or offshoring the work to India or elsewhere with low labor costs. The pros to these approaches include reduced costs and flexible bandwidth, but at the expense of quality control and work product consistency. Furthermore, with offshoring, export controls can come into play.

Insourcing involves law firms hiring unlicensed technical writers who effectively ghost-write patent applications under the supervision of patent attorneys and agents. These technical writers command lower salaries compared to licensed practitioners, so their hourly rates are also lower. One of the most challenging aspects of insourcing is often high turnover, which means that hiring and training are continuous activities. Quality control with insourcing is easier than with outsourcing because the work is locally supervised, but it can still be a significant project for an experienced practitioner to review and revise an application written by a neophyte. 

If we think about practitioner efficiency in terms of hours per application instead of dollars per application, then leveraging new technologies starts to make a lot of sense, specifically automation. Luckily, there is a wide variety of automation tools hitting the market. These include automated patent proofreaders, automated patent drafting, and other tools that shave off significant chunks of practitioner time per project. 

“If we think about practitioner efficiency in terms of hours per application instead of dollars per application, then leveraging new technologies starts to make a lot of sense, specifically automation.”

With automation incorporated into the preparation process, the time it takes for a practitioner to complete a patent application can be reduced by several hours. This means that that individual practitioner can process more applications per year and at a lower per-app cost. One law firm leader recently suggested that his firm, by using automation and process optimization, has individual associates generating over 150 unrelated applications per year. That is astounding. That’s about three applications per week per associate compared to one or less applications per week in traditional practices.

While automation also wins on quality control and work product consistency, there are limitations to today’s automation tools. For example, most require at least some adjustments to workflows. And some are not yet available in all patent domains or proficient at highly-complex patent projects. 

The patent bar may be in the midst of a perfect storm in terms of demographic changes and challenging economic realities, but the silver lining appears to be a burgeoning, tech-enabled patent workforce. Legal may be one of the last industries to be disrupted by AI and automation, but, at least in the patent ecosystem, the inflection point seems to be roughly now. As the trends discussed above play out, patent practices that fail to adjust course will struggle to remain profitable.

DISCLAIMER: The views and opinions expressed in this article are those of the author and (1) are not provided in the course of and do not create or constitute an attorney-client relationship, (2) are not intended as a solicitation, (3) are not intended to convey or constitute legal advice, and (4) are not a substitute for obtaining legal advice from a qualified attorney. You should not act upon any such information without first seeking qualified professional counsel on your specific matter. The hiring of an attorney is an important decision that should not be based solely upon Web site communications or advertisements.

Featured on:

Image result for law360

* Top ten most-read IP Law360 guest article of 2019.


Share this article

Top 10 Ways PatentTech is Disrupting Patent Procurement

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

A wave of new technologies is hitting the market with a specific focus on solving problems and inefficiencies in the patent procurement pipeline, from invention conception to patent issuance. This growing subset of LegalTech has come to be known as “PatentTech.” These new solutions are challenging the status quo of patent practice and becoming increasingly mainstream across the spectrum of law firms and in-house patent departments.

Two analogous technology disruptions in service industries include electronic spreadsheets in accounting[1] and computer-aided drafting in architecture[2]. In both cases, these disruptive technologies resulted in “a net positive for the industry with higher quality, higher efficiency, better access to services, and growth in the workforce”[3].

PatentTech solutions are wide-ranging, but generally fit into the following categories: document automation, process automation, and big data insights. Below is a breakdown of the top 10 ways in which PatentTech is changing how patent professionals and patent practices operate.

1. Invention harvesting: Ideation starts the patent process. Traditionally, patent attorneys often hold brainstorming sessions with engineering teams to identify potentially protectable inventions. Once identified, those inventions are assessed for business value, likelihood of eventual protection, and expected cost to pursue protection. This analysis then forms the basis for a comprehensive IP portfolio strategy. At least two companies–Legit.ai and Bright Marbles–are working in this space.

Automating this brainstorming process addresses strategic and practical aspects. Brainstorming automation can be informed by the prior art, specific competitor products, and the patent owner’s portfolio or the portfolios of their competitors, which can make the results highly-targeted based on business strategies. At the same time, automated ideation may offer more thorough mining via systematic and context-based approaches. Since the information gathered is in digital form, brainstorming automation should also force good IP inventory management. From a practical standpoint, being able to automate ideation eases friction with inventor/attorney scheduling and can even be performed asynchronously among inventors.

2. Invention disclosure: Once inventions are identified, the inventor must sufficiently convey the invention to the patent practitioner such that the practitioner can prepare a patent application that would enable one of ordinary skill in the relevant art to make and use the invention. One popular way of doing this includes having inventors complete a standardized invention disclosure form (IDF) covering the topics relevant to preparing and filing a patent application. Some companies, like ClearAccessIP and AppColl, have platformized this step in the patent procurement pipeline. These existing solutions seem focused on form-based approaches but are presented via a user-friendly web-based interface.

Indeed, anything that addresses friction in the disclosure process is a step in the right direction. However, invention disclosure could be one of the best areas of opportunity for future PatentTech solutions. For example, a semi-intelligent chatbot could certainly perform a more-than-cursory invention disclosure meeting that results in a completed, standardized IDF. The practitioner may still need to follow up with the inventor with some specific questions, but automating a majority of the disclosure process with a natural language interface accessible by the inventor anywhere, anytime would take a lot of pain out of the process and likely result in a higher volume of disclosures.

3. Prior art searching: Evaluating the relevant prior art is an important part of informing the decision of whether to file a patent and, if the answer is yes, determining the appropriate language and scope of the claims and specification. The goal of this exercise is to file patent applications with the highest chances of being allowed based on what is known about the prior art. Cutting out low probability applications and minimizing the number of office action cycles results in efficiencies enjoyed by all stakeholders–patent owners, practitioners, and the Patent Office.  

Several young companies, including Legalicity and DorothyAI, are automating patent search using AI. Automating this process, even if it’s just an initial screen of the prior art, can save hours of time for practitioners and professional searchers. In addition to speeding up turnaround on search projects, automation also makes it more feasible to do in-depth prior art analyses on a greater portion of a given portfolio.

4. Application drafting: The most time-consuming project in the patent procurement pipeline is drafting patent applications, which generally takes a skilled practitioner 20+ hours to complete. By segregating the judgment- and creativity-driven parts of patent preparation from the more mechanical aspects, automated drafting lets practitioners focus their talent on the truly high value-added components of the process. With intensifying cost pressure (see our prior post discussing revenue and cost trends in patent procurement), practitioners are turning to auto-drafting as a viable alternative to offshoring their drafting work.

A handful of rudimentary auto-drafting tools have popped up over the past year or so, mostly bespoke solutions developed in-house by drafting outsourcers and even by law firms in at least two cases. Fully- and partially-automated third-party drafting services, however, are now becoming available to the broader market. With Specifio‘s service, attorney-written claims are used to automatically generate a first-draft patent application with baseline 112 support. The idea, then, is for practitioners to bolster and finalize their auto-drafts ahead of filing with additional details and context around the invention.

5. Claim and spec proofreading: Sometimes small mistakes in patent applications can cost thousands of dollars to fix after the application has been filed. If a patent issues with certain errors in the text, it can even render the patent unenforceable or severely limit the scope of protection. As such, careful proofreading of the claims and specification is prudent (if not essential) ahead of filing the application.

While the rules are relatively simple, performing a thorough proofread by hand is tedious, time-consuming, and error-prone. Automated patent-specific proofreaders, like ClaimMaster and PatentOptimizer, have been around for a few years. However, the adoption of proofreaders like these seems to be accelerating. New AI-based proofreaders hitting the market, like BluePencil and PatentBots, are further evidence of how mainstream these tools are becoming.

6. Art unit prediction: Patent examination is still largely a human-driven process. And humans tend to be creatures of habit. As a result, the USPTO art unit to which an application is assigned once it’s filed can greatly affect a number of factors during prosecution, such as time to first action, number of actions before allowance, and, ultimately, the chances of ever being allowed.

In an effort to provide some control of the outcome, companies, like Juristat, provide an art unit prediction service. Coupling these insights with art unit analytics, such as that provided by BigPatentData, practitioners are able to adjust claim language to affect classification to improve prosecution outcomes and provide more realistic expectations to their clients.

7. Docketing: The patent process is a deadline-driven one. Missing deadlines can cost dearly in late fees and, for some deadlines, can cause the application to go abandoned. As such, docketing is one of the most critical aspects of any patent practice. Traditionally, the intake of USPTO correspondence and entry of corresponding deadlines is performed by a team of docketers. Beyond calendaring errors, the stress and monotony of these positions often result in moral issues and costly turnover.

Automation is becoming more and more prevalent in the patent docketing world. BlackhillsIP, for example, uses automation to augment its human docketers, while AltLegal provides a fully-automated solution. Both companies report cost savings and reduction in docketing errors. Building off of docketing, areas of opportunity for future PatentTech solutions include automated client reporting as well as schedule generation and workflow management for practitioners.

8. IDS preparation and management: Information disclosure statements (IDSs)–the mechanism by which patent applicants report known prior art to the Patent Office–make up another very tedious and error prone component of the patent procurement pipeline. Conventionally, this is often managed with spreadsheets, which is a problematic approach for obvious reasons. One of the earliest in the current PatentTech wave, SyncIDS, manages reference citations and relationships and generates ready-to-file USPTO-compliant IDS forms.

9. Prosecution: Preparing written responses to Patent Office rejections is one of the most time-intensive tasks for patent practitioners. With intense budget pressure, practitioners have found ways to maximize efficiency and leverage. Each office action response is unique, but there is definitely a rhythm to generating these documents. Traditionally, response “shells” are often produced by a paralegal or patent secretary. The shells include bibliographic info, claims in their current state of amendment, and an outline of the rejections in the office action. Shells take time and cannot have any errors. But due to the formulaic nature of these documents, shells are also great candidates for automation. Companies like ClaimMaster and TurboPatent offer automated shell generation as well as office action response proofreading.

When it comes to the practitioner’s part of preparing an office action response, examiner statistics are becoming essential. Examiner statistics are available from PatentBots, BigPatentData, and other sources. Practitioners can make data-driven decisions for things like “Should I appeal or file an RCE?” or “Have we successfully argued this particular issue in the past?”. One clear area of opportunity for the prosecution stage is automated claim charting, which could be leveraged during office action response drafting as well as infringement analysis, examination, and other aspects of patent law.

10. Strategic counseling/planning: One of the highest value-added parts of patent practice is strategic counseling to help solve business issues having an IP component. The analysis often considers various business objectives, likelihood of positive outcomes at the Patent Office, and the associated projected costs.

Patent filing strategies can be informed by a mapping between certain products (e.g., a competitor’s) and relevant patent claims (e.g., in an applicant’s portfolio and/or their competitor’s portfolio). Think of it as the opposite of a product clearance analysis where opportunities for patent protection are easier to identify. This is one of the myriad uses of ClearstoneIP’s platform, which also facilitates analyses like clearance/reverse clearance, FTO/reverse FTO, assessing licensing opportunities, enforcement analysis, competitor monitoring, etc.

Big data insights, like from BigPatentData, can help patent practice managers answer questions like “Should we prepare the client for lengthy prosecution?” or “Who at the firm is familiar with this technology?” or “Where is our prosecution practice most/least profitable?”. For in-house counsel, possible insights can inform decisions like “Should we pay for an examiner interview?” or “Which of our firms have handled this particular issue?” or “To which firms should we send more/less work?”.

More and more PatentTech companies are hitting the market and the rate is accelerating. Indeed, the companies mentioned above are just a few examples in the growing numbers of PatentTech. We are beyond a first-movers-only market with each area outlined above becoming increasingly competitive. Only a few years ago, it seemed like automation and big data insights were barely, if at all, on the radar of patent practice leadership. Now, in today’s hyper-competitive market, it’s dominating the conversation.

References:

  1. See Jacob Goldstein, How The Electronic Spreadsheet Revolutionized Business, NPR (Feb. 27, 2015), https://www.npr.org/2015/02/27/389585340/how-the-electronic-spreadsheet-revolutionized-business; also see Lisa Cumming, After VisiCalc Revolutionized Accounting In The 70s, AI Is The Next Big Breakthrough, BLUE J LEGAL (June 22, 2018), available at https://www.bluejlegal.com/blog/single-post/2017/12/05/after-visicalc-revolutionizedaccounting-in-the-70s-ai-is-the-next-big-breakthrough.
  2. See James A. De Lapp, et al., Impacts of CAD On Design Realization, 11 Engineering, Construction and Architectural Management, Issue No. 4 (2004), available at https://doi.org/10.1108/09699980410547630.
  3. The AIPLA/AIPPI/FICPI AI Colloquium Primer, AIPLA/AIPPI/FICPI AI Colloquium 2019 (February 2019), available at https://ficpi.org/_/uploads/files/AIPLA-AIPPI-FICPI_Artificial_Intelligence_Colloquium_Patent_ONLY_Primer.pdf.

DISCLAIMER: The views and opinions expressed in this article are those of the author and (1) are not provided in the course of and do not create or constitute an attorney-client relationship, (2) are not intended as a solicitation, (3) are not intended to convey or constitute legal advice, and (4) are not a substitute for obtaining legal advice from a qualified attorney. You should not act upon any such information without first seeking qualified professional counsel on your specific matter. The hiring of an attorney is an important decision that should not be based solely upon Web site communications or advertisements.

Featured on:

Image result for law360

Share this article

Patent Filing Behavior Among Resident and Non-Resident Entities

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

In our previous blog post, we explored how annual patent filings may be an indicator of optimism towards the strength of a jurisdiction’s patent system and its underlying economy. This article picks up the analysis to shed light on the effects of resident status on filing behavior over the past decade among the top five patent markets–US, China, Japan, Europe, and Korea–known as the IP-5.

The graph below shows the total number of patent applications filed each year in the IP-5 jurisdictions, with non-resident and resident filers aggregated together. At a glance, filing activity appears relatively stable with modest changes, at least compared to the rapid ascent and shear volume of Chinese patent filings.

Separating resident and non-resident filings shows interesting differences between the two categories of participants in each of the economies of the IP-5 jurisdictions. The graph below shows the number of patent applications filed by residents in each IP-5 jurisdiction.

In China, a hugely disproportionate amount of patent applications are filed domestically. In the last decade, 85% of patent applications filed with China’s National Intellectual Property Administration (CNIPA; shown in the figures with the former name, “SIPO”) were resident-based. From 2007 to 2017, China’s total number of domestic patents equaled 6.74M, which is roughly the same as the resident output of the U.S., Europe, and Japan combined. It would be interesting to see how this dramatic increase in filings has affected the number of those practicing patent law in China.

China has boosted domestic patent production though a number of initiatives. The country has seen significant investments in innovation, such as artificial intelligence. Also, there are patent-promoting policies, such as subsidized patent filing fees, rewards for patent filings, and tax credits related to patent output. China has assuaged examination guidelines for software and business methods patents and streamlined the examination process for rapidly expanding sectors, such as cloud, big data, and internet.

While Japan’s resident fillings have fallen, South Korea, Europe, and the U.S. have seen steady increases, suggesting a possible correlation with the positive economic activity of these jurisdictions. In the last decade, Japan’s resident patent filings, however, have steadily decreased by more than 20%, which may be explained by a sluggish national economy struggling to get back on its feet.

The graph below depicts the number of patent applications filed by non-residents for each IP-5 jurisdiction.

As illustrated above, the United States Patent Trademark Office (USPTO) is clearly the most popular destination for non-resident patent filings. For almost every single year in the last decade, more than half of the applications filed with the USPTO were of foreign origin. As the largest economy in the world, it makes sense that entrepreneurs and inventors want to file patents in the U.S. Due to the thoroughness of American patent law, the U.S. offers arguably stronger patent protection than other countries or regions.

Similar to the U.S., Europe has significantly more non-resident than resident patent filings, which may be due to the same reasons mentioned above. The U.S. and Europe have the most balanced production of domestic and foreign patent filings. The Asian market, on the other hand, is dominated by domestic patents. In the last decade, non-resident filings made up less than a quarter of total patent filings for China (15%), Japan (20%), and South Korea (23%).

In the plots below, resident and non-resident filings have been normalized to more easily compare the “shape” of the filing trends. This graph has been created so that maximum annual filings over the past decade are scaled to equal to one, while the minimum is scaled to equal zero.

It appears all IP-5 jurisdictions, except China, took a significant dive in 2009 with the USPTO, EPO, and KIPO hitting a local minimum. Japan’s steepest fall was from 2008 to 2009, but its patent filings did not bottom out until 2015. U.S. resident fillings also saw a dip in 2014, which is presumably due to the Alice and Mayo cases that cast doubt on the patentability of certain fields of innovation. It is not immediately clear what factors caused the volatility in Korean resident filings.

We see above that foreign filing entities maintained confidence in American patents through the economic crisis and the Alice-Mayo fallout with barely any fluctuations in steady year-after-year increases. Non-resident filings among the remaining IP-5, however, hit a low in 2009, but have mainly risen since.

The U.S. and China are the clear favorites in the international patent market, but in different ways. China is a mass-producer of domestic patents in part due to their favorable policies and incentives, while the U.S. has maintained a more balanced approach with a healthy influx of foreign patents. By integrating resident and non-resident factors into this analysis, we have gained deeper insight into the productivity and health of each jurisdiction, as well as the sensitivity of the overall market to external factors. Although Europe is the smallest producer of patents among the IP-5, they are still known for having one of the highest ratios of foreign-based patents, which is a testament to their strong patent system. It will be interesting to monitor the production trends of these different countries moving forward and whether they decide to take a quality or quantity approach in production.

So what does all this mean for practitioners?  We see above that diversifying clientele to include non-resident U.S. patent fliers and resident Chinese patent-filers could help weather economic downturns, which more often is followed by dips in patent filings. Furthermore, In order to chase Chinese patent work, practitioners’ best chances may be to develop business among Chinese resident filers.  

Data Sources:

  1. Increases in Innovation, Patent Boom Leads to Development in China http://www.ipwatchdog.com/2018/04/18/increases-innovation-patent-boom-development-china/id=95994/.
  2. Annual Patent Filing Data for IP-5 (2011-2017): IP-5 Statistics Report 2011-2018, Four-Office Statistics Report (2008 – 2010), World Bank Indicator https://data.worldbank.org/indicator/IP.PAT.NRES, and WIPO Statistics https://www.wipo.int/ipstats/en/statistics/country_profile/profile.jsp?code=CN.

Featured on:

Groen zwart.png

Share this article

The Relationship Between Annual Patent Filings and GDP

Share this article

By Ian C. Schick, PhD, JD, CEO & Co-founder of Specifio (first posted on blog.specif.io)

The top five intellectual property offices in the world, also known as the IP-5, dominate the global patent market. In 2017, the IP-5 received approximately 85% of the 3.17M patent applications filed worldwide. Showing no signs of slowing down, the IP-5 have increased their collective share of the global patent market by 10% in the last decade. The graph below shows annual patent filings for the IP-5 offices from 2007 to 2017.

A significant portion of the global growth of patent applications is attributed to China’s State Intellectual Property Office (SIPO; recently renamed to CHIPA). For example, the SIPO received 1.38M patent applications in 2017, which accounts for more than 43% of the global market share and beats the four other IP-5 offices combined. Following China and the U.S., the Japan Patent Office (JPO) ranked third with 318,479 applications, Korean Intellectual Property Office (KIPO) ranked fourth with 204,775, and the European Patent Office (EPO) came in last at 166,585. Asia, primarily China, South Korea, and Japan, has strengthened itself as the leading region with the highest patent filing activity, accounting for more than 65% of the world in 2018. While the SIPO doubles the second largest office, the United States Patent Trademark Office (USPTO), the U.S. has still experienced seven consecutive years of growth in applications. However, when matched with the strong growth of the SIPO, the USPTO, along with the four other IP-5 offices, have actually seen their shares of the global market decrease (see graph below).

This article explores whether there is any strong correlation between annual real gross domestic product (GDP) of the individual IP-5 jurisdictions and their annual patent filing trends. Indicative of economic well-being, GDP measures the total economic output of the goods and services of a country or region adjusted for the effects of inflation. The plot below shows the different GDPs for the IP-5.

Following the recent Great Recession, the U.S. and China have quickly rebounded and have shown no signs of slowing down with respect to economic growth, while the other remaining economies have responded differently. Since 2009, American GDP has been on an upward trend, overtaking Europe as the largest economy in the world. Europe’s GDP has fluctuated the most, hovering between 16 and 19 trillion. China has experienced tremendous growth in the past couple of years, while South Korea, on the other hand, appears to be relatively stagnant in the last decade. China’s GDP has grown more than 60% from 2007 to 2017, getting closer to Europe and the U.S. China and Japan, who were at similar GDP levels in 2007, went in different directions with China achieving significant growth in the last decade, while Japan remained at roughly the same level. Despite the recent growth in the US’s GDP, its share of the total GDP of the IP-5 has held relatively flat, as shown in the graph below.

To see how, if at all, annual patent filings correlate with GDP, we have calculated the number of patents filed per unit of GDP, which is plotted in the graph below. This figure is sometimes referred to as a “patent activity intensity” indicator. While variations in the number of patent filings demonstrate the size and level of development of the economy, this graph offers an alternative measurement that enables the comparison of the number of patent applications among countries with different-sized economies.

For a perfectly linear relationship between annual patent filings and GDP, we would expect to see flat lines in the graph above. Indeed, that appears to be the case for both the U.S. and Europe, despite the rising GDP of the U.S. and the sluggish GDP of Europe. In these jurisdictions, economic growth/contraction appears directly related to the rate of innovation measured by annual patent filings.

This apparent correlation disappears, however, when we look at Asia. South Korea started and ended the decade with the most patents filed per unit GDP, but it slowly decreased in general. China’s rate of annual patent filings has largely outpaced its growth in GDP as indicated by the rising trend in patents filed per unit GDP. Since Japan’s GDP took a downward turn in 2012, their patents filed per GDP has mostly risen even though their total number of annual filings has been in a steady decline over the last decade.

In summary, smaller economies tend to have more patents filed per unit GDP. This could be a function of GDP, but other factors, such as local labor costs for preparing patent applications, are likely at play. The U.S. and Europe show a strong linear relationship between annual patent filings and GDP. While not as stable as the U.S. and Europe, Korea’s patent filings also demonstrate a fairly stable linear relation to GDP. Japan and China, however, at least in recent years, have increased the number of patents filed per GDP. This may be due to confidence toward the respective patent systems and the value of patents for economic growth. In our next post, we’ll look at trends in resident versus non-resident patent filings in the IP-5 to shed additional light on the optimism toward each of the different IP-5 jurisdictions.

Data Sources:

  1. WIPO Facts and Figures 2018
  2. World Intellectual Property Indicators 2018
  3. Annual Real GDP Statistics for U.S., China, South Korea, Japan, and EU: https://data.worldbank.org/indicator/NY.GDP.MKTP.CD
  4. Annual Patent Filing Data for IP-5 (2011-2017): IP-5 Statistics Report 2011-2018, Four-Office Statistics Report (2008 – 2010), World Bank Indicator https://data.worldbank.org/indicator/IP.PAT.NRES, and WIPO Statistics https://www.wipo.int/ipstats/en/statistics/country_profile/profile.jsp?code=CN.

Featured on:

Groen zwart.png

Share this article