AI-Powered Patent Drafting with Deep IP
30:05 Share

Hello, everyone. This is your podcast host at the LegalTech Startup Focus podcast, Charlie Uniman. And I am very pleased to bring to the podcast FX LeDuc, a co-founder and the CEO of DeepIP.com.

A little bit of history we'll get into about the company's existence as an independent entity. But welcome, FX. Thank you, Charlie. Thank you so much for having me today. And it's going to be a shock to anyone that Deep IP is an intellectual property space, subspace, if you will, of legal tech.

A very important part has been for a long time, but with, of course, technology being as prominent all over the world and in so many ways, it's even more crucial than ever. So let me begin, as I often do on these podcasts, by asking the company's CEO and a co-founder to give us an idea of what Deep IP does,

what its functionality is and its benefits, and most particularly, as again, other guests have done, what distinguishes it from some of its esteemed competitors, what its secret sauce is. So tell us, FX, about Deep IP.

Yeah, thank you, Charlie. But I'll be very happy to speak about that. So basically, DeepIP, in a nutshell, is the trusted assistants that support patent practitioners to help them create better patents faster. And the kind of impact we are looking at is saving between 20 and 50% of the time.

Part of this time is basically reinvested into the most complex part of a patent, so mainly the claims, making in the end a more robust patent. -Let us get into the workflow of that. Perhaps that interests me when I went to the website, but go ahead. Sorry for interrupting. -As a patent practitioner,

your clients, the inventor brings an idea and you have to be able to rapidly take the most of the invention, better understand the novelty,

transform that into a claims, complete the detailed description before submitting it to your clients for review and filing it. And so basically, if you take a step back over the 20 past years, attorneys have been asked by their clients to be able to work faster and faster.

They used to have something like 40 hours to draft a patent on average. And today they have something like 20 hours to be able to do the work. And so they have found the industry has been under pressure for long years now. And they were basically extremely happy when the technology started.

has emerged because there was a hope to find out ways to be more productive and to get back control over the workload and to be able to deliver quality work. And so DeepIP basically is really an assistant that is fully embedded within your Word documents

that you can open where you work, so basically Word. There is no disruption switching back from a web application to your Word documents. You can directly dialogue with it within Word.

It's a very iterative drafting process. Basically, the one-click button to get a patent doesn't work. There is no AI that is good enough to be able to do that. The way we design the solution is to make it super iterative so that the attorneys keep control over the AI and guide the AI in the right direction.

And thanks to that, you will save between 20% and 50% of the time on the overall drafting time. And what we see is that attorneys tend to reinvest this time by making a more robust description, making more robust claims, and in the end delivering a better product to their clients. And I recall seeing that, looking at your website,

One of the newer developments coming from Deep IP was what to do if I, and I practice law, as many of my listeners know, but I was certainly far from the intellectual property patent side of things. But when the examiner, I'll call that person, comes back with comments or reactions, that's an area too, am I correct, where Deep IP can assist the practitioner?

Yes, definitely. We started with patent drafting support and we developed a second product, a second module dedicated to what we call office actions, which are basically what you describe, the letters coming from the PTO with rejections saying that your patent is actually not novel or there is a prior art or the idea is not patentable at all.

whatever the kind of production. And so on average, there are two office actions per patent. So if you think as a law firm, law firms have a huge amount of section to handle. And what DeepIP does, it basically it is connected with a USPTO file wrapper. So it fetch the entire set of documents.

It identifies where are the rejections and what are the kind of rejections. It is able to split up the claims to identify by claim elements which of the rejections from the PTO are weak.

and it helps find out arguments to be able to address these rejections in order to be able to provide a robust response to the PTO and get your patent granted. And I think... Go ahead, please. I think what is important to have in mind is that these technologies, so DeepIP supporting the attorney,

It's not nice to have. I think it is becoming more and more a must-have for any IP law firm to be able to address

which are getting equipped on their side as well. So basically all PTOs, all major PTOs in the world are developing their own AI technologies to be able to read patent application and to draft automatically rejections. So basically detecting issues and

and to push that back to the attorneys. And so I think one of the main transformation which is happening is that AI is going to be everywhere on both sides, on the client side, on the attorney side, on the PTO side. And firms have to look at it as a kind of urgency to get equipped, learn how to use this tool.

And in order to be able to keep up with the way the entire industry is evolving. Well, I'm going to get on the hobby horse that I've ridden before on this podcast and say to you and our listeners, FX, that it is fast coming upon us where a law firm can't adequately represent clients.

and zealously represent clients as often it's described in regulations governing the way lawyers practice unless they have these tools, let alone stay competitive in the market. And now, and actually from among the 60 some odd podcasts that I've done to date, you're the first to mention that the regulators themselves, it makes perfect sense as long as they have the budget

are themselves using these technologies, most particularly AI, generative AI in some way, shape, or form. So if the law firm is not equipped to use AI and meet the regulator on the AI field, they're going to be left behind. So

Being the proponent of technology's use in the law that I've been for a very, very long time now, this is even a greater indication that law firms can't look at technology and, of course, now AI and say, oh, it's something that we can ignore and passing fad and all that. We're way beyond that point. Before we hit the record button, I was talking to FX a little bit about the –

The latest and greatest flavor of AI that people are buzzing and talking about, and that's agentic AI. Those of you who are listening in the New York area should be on the lookout. I'll put in a shameless plug for the New York Legal Tech Meetup. It's an organization that I'm involved with. It's not a profit organization.

making organization, but it's a wonderful place to meet up in the city. And their next event is planned for some kind of panel discussion of agentic AI. To me, that means the AI is not just generating language, but it's also going step by step through workflows and perhaps even making decisions. What about that aspect of AI that

in IP work and at DeepIP? Yeah, I think it's a great question. So first of all, perhaps coming back on the definition of agentic, I fully agree with what you shared. So basically, what

Before Adjantic AI, we were speaking about large language models. And the main way to interact with these models were basically a chat room, as you, I believe, all experienced using ChatGPT. And what is new with Adjantic AI is that agents are capabilities, AI capabilities, which are able to

move things across the workflow step by step, taking potentially decisions or asking humans to provide feedback at a certain step when there is a decision the AI is not able to make before going further down into the task on which they are working. So basically it's a kind of autonomous or semi-autonomous

AI agents able to do a broader part of the workflow. We were making a webinar earlier this year and I was sharing that one of the key developments, actually I see two key developments for AI in 2025. The first one is obviously the emergence of agentic AI. Basically being able to do more in a more autonomous way.

and this is what we are obviously developing at DeepIP. The second one being what we call personalization. Basically, large language models at the beginning had no memory. They were not able to memorize previous interactions you had with the AI and to adapt based on these previous iterations.

Now, memories have increased and it is possible with these technologies to be able to personalize the AI to your specific style. And so, agentic and personalization are the two key evolutions, I would say globally in the AI space. And obviously we are implementing these strategies within DeepIP. What does it mean? It means that

the way to draft will be a bit less iterative because the AI will be able to do a bit more about the workflow. As these capabilities are evolving, and as you know, it's evolving very, very fast, the broader part of the workflow, this technology will be able to make by themselves. I think it's a very, very exciting moment.

And regarding personalization, to be very concrete, as an attorney, everything is about your style. You have a style to draft for every client, for each client, for each domain, and potentially each attorney has its own way to draft patents. And so basically what we do

A law with DeepIP is you will be able to use previous applications which have been published, load them into DeepIP, and our AI is able to understand your style, extract the pattern of styles, and we use it as a fine-tuning layer

so that you can leverage this style to draft future applications for these specific clients/domain or technology. Amazing. And yeah, I often find myself thinking about what it would have been like had I continued to practice and had the benefit of these developments technologically. But having a personalized AI, meaning that it's not just

an assistant generally for IP lawyers or M&A lawyers, but it's your AI who knows your preferences for the style with which you draft things, the issues that you frequently think and come across as the most important in your area of practice, and

And I couldn't help but think, as we were talking about agentic AIs, that someday the PTO office and the law office will just let their AIs talk to one another for a while. And then the humans will get involved and be in the loop and settle the matter. It also got me to thinking, and we're going to take a bit to talk about this.

what I had described as a niche market to FX, but before we hit the record button, but I think he corrected me. We'll get into that. You know, there are so many other areas, immigration law, and I have a podcast coming out soon talking to a wonderful new startup in that field, but also, for example, Securities Exchange Commission practice in the U.S., where you have these routines and you have...

regulators that are probably themselves arming themselves with AI. And I have to look a little bit more closely to see if something along the lines of what you're doing in the IP space is being done. I'm embarrassed to say, I don't know off the top of my head in the SEC space where at one point in my career I practiced, but especially caught my attention when you say that the PTOs

are going to get AI behind them because I wonder what the SEC, subject to budget constraints, will be doing with AI. Let's talk about the market that you address. I had sort of fashioned in my own mind a picture of your selling marketing Deep IP to law firms, but I think

I stood to be corrected there. It's a broader market than that, is it not, FX?

Yeah, the way we look at the market is basically, so we are not looking at it from the legal point of view, but more from as an industry. And so IP is basically a two-sided market. On one side, you have companies which are developing their technologies, where the inventors are, and they are developing their portfolio of patents.

and they collaborate with attorneys which are the outside consuls for some of the part of the workflow so typically sometimes pre-patent consultation work like FTO analysis for instance freedom to operate analysis for people which are not familiar with IP then you have the drafting then you have the prosecution potentially you have the litigation

But on the corporate side, you have also a large part of the workflow, which is basically understanding from all the inventions which are proposed by your R&D teams, which ones are worth patenting because it costs a lot to patent.

Which technology do we want to bet on? Which patents should we stop? Should we prune? So there are a lot of steps of the workflow which are also on the corporate side. And so we look at the market from this point of view. So basically a two-sided market.

and we ambition to develop basically the AI operating system for IP serving both attorneys on one side and corporates on one side, following the entire workflow from inventors to pruning. I hadn't realized that there was that in-house enterprise side to the market that you can address as well.

With that said, the fact that you have patent attorneys in private practice and then you have in-house attorneys and you have the inventors in the IP departments at the enterprise, how do you reach out market? Is it conferences? What are the most significant ways in which you get word out about what's going on at Deep IP? Yeah. So we...

I think it's a very interesting question. And so before starting, we wondered if we should start by addressing the corporate market or the attorneys market. And we decided to address first the attorneys market because this is where we felt the most urgency to create, to bring productivity, basically. So to bring a bit of relief in the practices, to help them be more productive,

and be able to restore a bit their margins and be able to provide a better work to their clients. And this attorneys market is quite concentrated. So if you take the US market, there are 700 high pillow firms. And if you take the top 100 high pillow firms, they represent 50% of the market. And so it's a relatively niche market. These people...

attorneys, they meet very regularly at some large events like IPO, AI PLA. And so obviously we are there and we speak about what we build and how we see AI transforming the field. We also develop very strong relationship with our clients. So we work with some of the largest IPO law firm in the US, Schwegman, Cantor-Colburn, Greenberg-Traurig, to name a few. And so our clients speak

about what we do together. And so we do webinars and we share these webinars to the community to help other firms understand what is happening in the markets. And so what are the early adopters doing, the way they think about AI, the way they see the impact. So basically what does it change in the way they operate?

how they successfully designed adoption plans to make sure that these tools are effectively used by attorneys so that they really see the benefits of it and build the ROI. And so we basically do a lot of shared events with our clients to let them speak about their experience. I think it's always important

more convincing to listen to one of your peers rather than the vendor, which tends always to say things which are not always trusted.

And finally, we create a lot of content. So we publish a lot of articles that we share with the community to help educate attorneys on what it means to use these tools, how to use them, what might change and best practices.

Basically, the idea is to create a lot of value and to share it with our market and they come to us. Yeah, I think that's so important to realize that if you can target webinars, you can target thought pieces.

You can cooperate with your clients and jointly present to others in the area. You know you're not wasting your marketing dollars because it's not broad consumer spending. It's not just reaching out to the market at large. You're focusing your dollars on just the people to whom you want to sell the product.

And, you know, I've often said to startups that while it does take time and it does take effort and some skill to win the title of one of the thought leaders in your field, it's not a big dollar outlay often, and it packs a big punch as far as marketing goes. So if you're fortunate enough to have a target market where, you know, you can really concentrate on

on not only conferences and face-to-face meetings or virtual meetings with a tight-knit group of attorneys or enterprise people, but also where you know they congregate virtually to read things and discuss things in webinars. Take advantage of it.

Don't hesitate even to bring on a hire who can learn the field right about it I think it's going to pay back your your marketing spend many times over I Think when we talked before the record button was hit you you mentioned Asia do you

sell to market the product in Europe and other parts of the world as well as the US or is it principally a US focus? We started the company in the US for the US market and today I think the US still weighs something like 75% of our total revenue so we are still heavily

focused on the US market. But as you can hear, I'm French and so I am so very much rooted in Europe. A part of the company, basically our engineering team is in France, in Paris, which is a thriving AI ecosystem. And so we also...

developed the activity in Europe. And today we work with some of the largest IP law firms in Europe as well. Plasro, which is number one in France. Giacobacci, which is number one in Italy, for instance. But what we see that basically, Europe is always a bit laggard compared with the US.

So the market will really take off, I think, this year, basically. And regarding Asia, it's very interesting. We're not proactively selling in Asia, but we see a significant increase in demand coming from China mainly and Japan as well.

I believe there is also an opportunity in Asia and it's a matter of time. We are still a young company. We started the activity less than two years ago and we are structuring things step by step to make sure we are able to properly address each market and deliver the level of success we would like to deliver to our clients.

Again, not ever having practiced in the IP area, still it rubbed off on me that so much of what has to be done involves making sure that you've secured your patent protection and other sorts of IP protection if you're a U.S. company, not only in the States.

but around the world, so it's wonderful to hear and not surprising to hear that Deep IP is giving your background in Europe and the fact that you're here in the U.S., of course, operating yourself from New York means that you've got that ground cover, but also that China and Japan and others in Asia are going to be paying attention to Deep IP. It used to be that...

When I was practicing, we looked to London firms as being among those that led in the use of technology and the law. But I think finally the US is not only catching up, but becoming one of the leaders over the last 10 years or so in legal tech. And I enjoy hearing it said that we are a little national pride.

So I want to thank you, FX, for joining me. It's been a pleasure to talk, to learn about Deep IP, get an idea of what AI means and that part of our legal tech vertical. If people want to learn more, is it deepip.com to learn more online about the company?

Deepip.ai. Oh, .ai, okay. You know, I type it in and I see it and I click and I didn't realize what the URL was. So deepip.ai. All right, everybody, make note of that.

Thank you so much, Charlie, for having me. It was a pleasure to chat with you and to share a bit what we build at Deep IP. Yeah, no, it's a pleasure. It's mine. Since I live not far from New York, I look forward, as I say, at the end of virtually every one of my podcasts, meeting you in real life when I get into the city. And maybe we'll cross paths at Legal Week next week because a lot of people are going to be walking around. And I'll be at that conference. Again, thank you and goodbye.

Continued wonderful success with Deep IP. Thank you so much. Thank you for listening to the LegalTech Startup Focus podcast. If you're interested in LegalTech startups and enjoyed this podcast, please consider joining the free LegalTech Startup Focus community by going to www.legaltechstartupfocus.com and signing up. Again, thanks.