Listen on Apple Podcasts, Spotify and YouTube
Welcome to AccelPro IP Law, where we provide expert interviews and coaching to accelerate your professional development. Today we’re featuring a conversation with Evan Zimmerman, CEO and Co-founder of Edge.
Edge is a Y-Combinator backed tech company helping patent practitioners draft patents and disclosures. Zimmerman discusses building a patent assistance platform, the patent application process, localization for patent work and more.
“The people who get into IP Law are some of the most innovative people in the legal profession. And yet, they have some of the worst tooling,” Zimmerman says.
”Almost 20% of patents are rejected with an office action reason that includes an error. We realized there was a need to create patent software to help unlock more of the innovation and help make patent practitioners’ jobs a little easier.”
Listen on Apple Podcasts, Spotify and YouTube.
Interview References:
Evan Zimmerman’s LinkedIn profile.
9:56 | Novelty 35 USC 102 Obviousness 35 USC 103. USPTO.
14:59 | 609 Information Disclosure Statement [R-07.2022]. USPTO.
17:14 | Alice v. CLS Bank: United States Supreme Court Establishes General Patentability Test. (August 2014). WIPO Magazine. World Intellectual Property Organization.
18:55 | McRO, Inc. V. Bandai Namco Games America, 837. F.3d 1299 (Fed. Cir. 2016).
Supplemental Materials:
Zimmerman’s venture capital firm: Jovono
Edge’s company blog: Nonobvious.
TRANSCRIPT
I. THE PATENT APPLICATION PROCESS
Neal Ungerleider, Host: For our listeners who aren't familiar, can you tell us a little bit about Edge and how it works?
Evan Zimmerman: Edge is a platform that helps practitioners work with their patents. Specifically, today we focus on prep and pros, and we do that by having a combined interface and assistant that helps with the disclosure process. We help your inventors get more out of the process without having to have as many meetings, and then we help you, the practitioner, actually draft the patent through a combination of user interface features such as reference tracking and figure labeling, and AI features such as having the assistant help with writing a detailed description, creating a table, or even helping you with the claims.
NU: What inspired Edge as a product, and what gap did you see in the patent application process that Edge can help with?
EZ: My dad's an inventor with over 30 patents. The thing that I found kind of interesting was that with intellectual property, and especially patents, it's about innovation…it’s about progress.
The people who get into this are some of the most innovative people in the legal profession. And yet, they have some of the worst tooling. I experienced it myself. The companies that I work with were all patent-backed. I even started a venture capital firm. We sold a company to Walmart and it was in part because we helped them get patents that Walmart wanted.
What I saw was everyone writing in Microsoft Word and using PowerPoint to label the figures. You actually see this in the numbers too. Almost 20% of patents are rejected with an office action reason that includes an error. We realized there was a need to create patent software to help unlock more of the innovation and help make patent practitioners have a job that's a little easier, a little more fun, and a little more effective, so they can focus on just doing their job and running their firm and the creative parts of being a patent practitioner.
NU: Can you walk us through the patent application process from drafting to initial submission to approval? And what can our audience do to make their workflow easier?
EZ: For the patent practitioners listening, this will be old hat, but for the copyright and trademark, this might be a little newer—you first start with someone who comes to you and tells you what they've invented.
If I don't tell you what I've invented, you won't know what to patent.
It's actually a lot more of a dance than people might think. A lot of inventors are not necessarily the best at describing their ideas. Sometimes they'll have superfluous things that they add in. Sometimes there will be variations that they don't realize are there.
One that I loved was a patent practitioner who told me that he and all of his inventors will come in and mention the language that they wrote their original software in. That has nothing to do with the invention; it's a little more abstract.
You actually have a lot of conversation that goes on where the patent practitioner takes notes and summarizes those notes. This can be hours in and of itself, both for the practitioner and for the inventor, and it’s something no one really wants to do.
Once you go from there, you've learned what the invention is. Then you start drafting. Most people start with the claims first, but that's not necessarily a rule. We even have customers who do the claims last.
From there, you usually go through an iterative process where you'll fill in the pieces of the patent. Patents are about claims, which is technically what you get from the patent. It's your legal rights. Think of patents as a very strictly defined property right.
I think of intellectual property as a spectrum, where the stronger and harder it is to obtain, the more strictly it's defined. Patents are on the extreme end of that spectrum with the strongest enforcement. It's strict liability. In exchange, it is the most difficult to obtain and has the most clearly defined rights.
Then you have the specification. The specification is the long part of the patent, and it explains what the claims are—what the scope of them would be.
There's a lot of jurisprudence on how to interpret specs, or detailed descriptions, as they're called. We'll set that to the side for now. It's typically iterative. You'll go through it, you'll find things that you like and don't like, and add embodiments.
You'll usually engage with the inventor at least once to see if you have things right and if they like what you did. Sometimes there are improvements that come along the way. Maybe they invented something in the middle that you’ll want to add extra claims in for or put extra details in the description. Then you file.
Note that you may file a provisional or a non-provisional. Provisionals are basically placeholders. Patent offices have different national jurisdictions that vary in how strictly they interpret the relationship you have to have between the provisional and non-provisional, although everywhere it is tending stricter. The EU is one of the strictest.
From there, what you'll do is file. If it's a provisional, you eventually have to file non-provisional or it dies. Then, you go through the prosecution process. If you're lucky, you'll get granted without an issue, but if it's not, you may wait up to two years in some groups, even three or four, to hear from the patent office, and they'll say what they agree with and what they disagree with. Those are called office actions.
You'll either accept or reject it. If you reject it, you have to negotiate with the patent office and make your case. More commonly, you'll accept some, reject others, and try to compromise. Then you will edit the patent based on that, and you'll have an issued patent that you can then start to enforce.
I know that's a detailed discussion, but that's how patents work, at least on the prosecution side prep and pros.
—
II. PATENTS AS A HIGHLY REGULATED INDUSTRY
NU: Patent law is a highly regulated field. What obstacles did you encounter making software specifically for the patent application process that seem specific to a highly regulated field?
EZ: Typically, what people have focused on before Edge was process automation stuff—things like helping manage the submission. Even then it was typically rules based, not a ton of stuff for the writing of the patent itself.
Patents are definitely a regulated industry, like any legal industry, in the sense that they’re highly legalistic. They’re very specific and the lawyers themselves have a lot of stakes. It's different than, for example, the medical industry where you have things like HIPAA. We don't have to comply with that.
Our customers are all attorneys, whether they're in-house or outside counsel, they are attorneys. They have ethical obligations from the bar, and those include duties of confidentiality, duties of care, and duties of performance.
First of all, and this one's kind of obvious, our software has to work. We can't get people in trouble for using our software.
The more important obligation is confidentiality. Confidentiality is something that's really key. You don't want to break your privilege, and even more importantly for patents is that a break of confidentiality can destroy the patent.
If you have a public performance, even inadvertently, you basically create prior art, and prior art will invalidate your ability to patent under a novelty 102 rejection. Inadvertent prior art starts the clock. You want to be in control of that timeline.
What that means is that we have to have a lot of cybersecurity and need a lot of privacy protections in place.
A really important thing that people underestimate is there's a difference between what you do and what you make people understand and feel comfortable with.
A huge part of our job is not just having cybersecurity—we have really good cybersecurity—but engaging with ethics councils and talking to GC’s. We put together presentations that explain in plain English what we do and why it protects people. We sometimes even connect it back to bar opinions issued on the ethics of using various types of tools and what types of privacy requirements they're supposed to look at.
We really just make it easy. I think that it is always the job of the vendor, not the customer, to make things easy. If someone has a question, why should you have to do your own research? We should be doing the legwork for you so that you can come in confident. Frankly, we’ll come up with the same answer you probably would on your own, just with a lot less work.
NU: Do you find yourself having to address concerns about potential job displacement or the ethical implications of using automation or AI in the legal industry? How do you address those concerns?
EZ: We hear that a lot of people are trying to figure out what this means for junior employees. What does training look like? How much responsibility do you give people? Do I need the same number of paralegals that I needed before? And so on.
One thing that makes us different than a lot of the other companies is that we aren't ‘Press a button, get a patent.’ We are a task based tool that helps with the writing process. The way that we address these issues is by engaging with customers—hearing what they're doing and working with them on programs that can actually benefit the entire legal community.
Over time, I wouldn't be surprised if we have training programs that we give away to practitioners who adopt our software. This could even be a paid service where we have ongoing training sessions and tune-ups.
It's like anything else when computer technology first started to become a thing.
Do you know why Lexis and Westlaw used to charge per search (and are moving to a subscription model)? It's because it used to be a job. There were contractors paid to go to legal libraries and pull information, and they charged per search. That's a job that doesn't exist anymore.
But it’s not like the legal profession has gotten smaller. The US has more lawyers than it's ever had. What's happened is that what people do has shifted. The types of patents they do are going to be somewhat different. It's going to be less run-of-the-mill patents and more crown jewel patents. They will spend more time on patent strategy rather than pure drafting, although it's still going to be a critical part of their business. It's going to be a huge part of what brings people in to trust them to do the patent strategy.
It's less about displacing people and more about displacing certain work.
NU: At this stage, Edge is creating documents for USPTO filings. Are you looking to expand applications to other regions and jurisdictions? If so, how do you ensure that generated templates comply with relevant patent laws and regulations in different jurisdictions?
EZ: We're definitely looking at more places. We already support close to ten languages. The next step is making sure that there's some localization in terms of the patent regulations.
Typically, what you'll see is that people file a PCT and then file in these national jurisdictions. The actual patent tax doesn’t undergo major revisions for the first thing. That's precisely what the PCT doesn't do. The initial drafting is not so different based on where you start. It's really the prosecution that changes a lot.
One example is that in the EU you have two-step claims. However, if you talk to EU patent attorneys, almost no one starts with the two-step. It's only if you get a complaint from the patent office, the EPO or one of the national jurisdictions, that you then go and do a two-step process.
Another example is when you do a claim limitation that is supported by the spec. In the US, you can say some kind of limitation and they'll usually accept it. Whereas the EU is very strict about whether you have the support for it in the initial draft.
The truth that really comes up is in Stage Three, where you're going back and forth with the patent office. That's really where you see more of these differences.
I will also say that process automation is the other place where you see big differences. For example, in the US you file your prior art with an IDS. That is not what you do in the EU or in Japan. They have different names for those forms. But that's more pure process automation. That’s more about things like ‘Do you want to have more detailed specs?’ as you sometimes see in Europe.
That's a user preference. If you want to add more embodiments, add more embodiments. That's not necessarily a regulatory difference—that's more of a general practice and a difference that you'll see between practitioners. We have customers who do completely different levels of detail. This is based on the nature of the things their customers are patenting, based on their personal preferences, and what they've seen get approved and disallowed.
NU: How does Edge integrate with or complement the work of patent attorneys or IP law firms?
EZ: A key part of what we do is that word that you use, complement. We aren't a people-replacer. We are a people-augmenter. Edge helps by taking away some of the tedious work like describing figures or embodiments.
The next thing is avoiding errors. Everyone's heard these horror stories of AIs filing briefs and making up fake cases. You don't have that here. The fundamental reason has to do with the nature of the application. Patents are less about conjuring things and more about defining applications and descriptions within the scope of the disclosure.
The last thing is we help with customer satisfaction. Customers want things done faster. They want it done easier. They really like the disclosure tool, because people don't like feeling like they're paying for describing what they're doing. They want to avoid mistakes, which good disclosures help with. Good inputs make good outputs. That's true in the patent world as it is in anything else.
Ultimately, what we see is that if you're a patent practitioner you're in the business of making satisfied clients and helping them achieve their business objectives in their innovation through intellectual property. If we can help you create more satisfied clients, both through a better client experience and by better serving the core need the client has, we’ve fundamentally helped you with your business. We've helped you become a better practitioner and a better patent practice, business owner, or business operator.
NU: How should lawyers and other legal professionals think of AI as a tool in terms of approaching workflows and project management?
EZ: I'm a really big proponent of task-based AI and task-based computing, as opposed to agentic computing. I think it's not totally there yet. Agentic means that you can tell it to go off and do a task and it will not only do it, but make judgments about what to do next.
The thing with these AI tools—first of all, there are a lot of them—second of all, you should really be taking a serious look, because one of the issues that legal tech had in the past was with workflow managements. You can make it easier to search, that's a huge deal; Then, you can make it easier to do form automation, that's great. But what's next?
What AI does, which could have never been done before, is help you with the actual content. Things that require you to do more than just fill out a form. That is a huge deal which was not possible before. If you take patents, it can even help you find the prior art, not just fill in the IDS and help you draft the patent— not just help you file it.
It's the same thing for a lot of professions. If you're in trademarks, can AI help you do a trademark filing? If you're doing copyright, can AI help you find potentially infringing works? Or help you cut the time to determine, at least at a first cut, whether there's an infringement?
Think about how it can help you run your practice. I'm seeing a lot of AI based timekeeping tools, for example. I would encourage practitioners to think and consider how purpose-based AI is better; using a tool that's meant for lawyers rather than just a general tool. Because you don't have to learn how to prompt engineer. Practitioners are busy people. Spending time learning to use a tool can be a good investment, for sure. You have to learn how to do some stuff, but things like prompt engineering, that's not a good use of your time.
The last thing is procurement. This is something that is new to law firms in general. What did you need to get started? You needed people in an office and maybe a word processor. That's it. Now, you're looking at a world where practitioners are going to be paying for potentially five extra new pieces of software, maybe even more as time goes by. Procurement is something that your clients are very used to. They're used to buying software. If they're in hardware, buying things like scalpels, or power generators, that’s the thing that patent practitioners and/or lawyers in general are not used to.
Think about what your procurement process is like and what are the best practices: actively searching, doing quarterly reviews of problems and potential solutions that could fix those problems, doing bake-offs, and those kinds of things.
I think that the confidentiality and ethics review that we talked about earlier will be part of the lawyer procurement process. That's not the same thing for other industries. I think that you're going to see other things, but they're going to be more based on the practice area.
To summarize what we talked about, like task based AI, think about what you're using it for. Consider it as part of the whole management of your practice and what you do for your job, not just for the workflow automation, and procurement. Think about how you're going to get tools, and not just the tools that you're going to use.
–
III. VENTURE CAPITAL AND PATENTS
NU: You're also involved in venture capital firm, Jovono. Can you tell us a little bit about how patents and intellectual property intersect with the venture capital world?
EZ: Patents are really interesting because of the Alice decision. It's been pretty difficult to get software patents. That makes them, when you can get them, more valuable.
For Jovono, the key thing to ask is, How can patents give you a competitive advantage? How can you use them to get protection, given that there are long patency times for patents and even getting a review, let alone an issuance.
The other thing to think about is, How does this help you with your terminal value? We invested in a company called Zeekit, and one of the main reasons it got acquired by Walmart for over $200 million was that it had software patents that were really good. Patents can be a really big part of the game.
I will also say that not all venture capital goes to software companies. First of all, you're seeing a lot of money flow to AI companies now. AI patents are easier to get than regular software patents. Certainly, that's a thing you're seeing a lot more of in startups. When you're talking about things like biotech or hardware, patents are essential.
NU: From your perspective, why are AI patents easier to get than other kinds of software patents?
EZ: First of all, if you look at the numbers, software patents don't have very high allowance rates, and that's even accounting for the fact that, post Alice, there's a selection away from filing for patents in the first place. The true idea to patent ratio is much lower than even the allowance rates say.
With AI, you have more allowance because the Federal Circuit started issuing opinions like McRO, allowing you to have some notion of inventive step or inventive application. You could take software, and you couldn't necessarily patent an algorithm per se, but you could apply algorithms that had certain characteristics. Humans could intervene and modify a parameter or software that's part of a broader application.
So much of AI is for these broader applications. There have been years where nearly half of all AI patents were machine vision applications for various medical industrial applications.
Similarly with AI, you have these use cases that tie into human intervention in some way. It just fits into this little narrow corridor that the Federal Circuit has laid out.
As an additional factor, and I'm being very blunt here, I think that after things like the Amazon patents that a lot of examiners have been very sensitive to the idea that they issued a lot of patents.
AI is so much newer. It's so much more innovative and transformative. Examiners look at this and say, “This is a real innovation.” You do have these requirements of subject matter, eligibility, novelty, and non-obviousness. Those are the core causes of rejections. You can put those all into an intuitive bucket of ‘is this real or is this a joke?’ And it is humans making these decisions.
I think ‘is this a joke’ is a thing that examiners really consider, even though they won't use those words ever. If you have a guy who says “I want a patent because I put a button somewhere,” a lot of examiners are not going to be sure about that. But if you say “I have this really cool machine learning algorithm that can be used in a clinical decision support system that helps determine whether to help someone with lung cancer use a particular treatment,” they're going to say that seems like a big deal.
NU: Can you give the AccelPro community an example of a time when a peer helped you with a difficult challenge during your career?
EZ: Y Combinator has been really transformative to our company. It's the world's top startup accelerator and we had a lot of peers who recommended us, did practice interviews, and really helped us get up to speed. This helped us not only get in, but hit the ground running and make the most of the program.
That’s the thing that people most misunderstand about Y Combinator. It's not a magic wand that makes your startup successful. It's a lot of resources and people who care about you and help and offer community, but you have to make use of it. It's a ‘what you give is what you're going to get’ situation.
Listen on Apple Podcasts, Spotify and YouTube.
This AccelPro audio transcript has been edited and organized for clarity. This interview was recorded on April 3, 2024.
AccelPro’s expert interviews and coaching accelerate your professional development. Our mission is to improve your day-to-day job performance and make your career goals achievable.
Please send your comments and career questions to questions@joinaccelpro.com. You can also call us at 614-642-2235.
If your colleagues in any sector of the IP law field might be interested, please let them know about AccelPro. As our community grows, it grows more useful for its members.
