The patent attorney profession is, by definition, built around protecting innovation. For roughly 100 years, it has not had to practice what it preaches. That is changing fast, and AI patent drafting and prosecution tools are at the center of that shift.

The founders of DeepIP and PatentMaker came together for the first time recently for a live panel discussion on what their acquisition means for German IP firms, and where European patent practice is heading. The conversation covered the state of the German IP market, the strategic rationale behind the acquisition, what the combined platform delivers today, and what AI-native patent prosecution looks like in 18 months.

Speakers:

• François-Xavier Leduc, Co-Founder & CEO, DeepIP
• Édouard d'Archimbaud, Co-Founder & CTO, DeepIP
• Dr. Matthias Hofmann, Co-Founder, PatentMaker
• Celia Jinlan Wei, Co-Founder & Managing Director, PatentMaker

Special Guest and Moderator:

• Johannes Ernicke, German and European Patent Attorney; Founder, Patenza

Read on to discover the most important things they got right.

‍

‍

The Profession That Never Had to Innovate, Until Now

"The patent attorney profession is all about dealing with innovation, but we never really had to innovate that much ourselves over the last 100 years,” Dr Hofmann opened. “Someone who learned this job 30 years ago or longer can still do it exactly the way he learned it, until just recently."

The workflows that define patent practice today—patent drafting, office action responses, prosecution strategy—were not designed for AI, as Dr Hofmann observed. They were designed for a world in which skilled human labor was the only option. In his view, AI does not just speed up those workflows. It makes the underlying model visible in a way it never was before.

For European firms in particular, Leduc noted that the reckoning is arriving later than in the US (the American market is 12 to 18 months ahead in terms of AI adoption and maturity), but the shift is arriving with the same force. 

Ernicke added that the German market faces a particular cultural tension: a deep commitment to stable processes and high reliability on one side, and the need for speed, change, and the courage to invest on the other. "You can't only focus on one of them," he said. "You have to find a good balance."

‍

How Should Law Firms Approach AI for Patent Drafting and Prosecution?

The most common mistake firms are making right now, according to d'Archimbaud, is treating AI adoption as a procurement decision: selecting an AI patent tool the same way they would choose any other piece of software.

"This transformation is not only about technology and software and buying software using AI," he said. "It's deeper than that because AI will deeply transform the way we work: the workflows, the touchpoints, the way we collaborate."

Leduc went further, arguing that firms treating AI as a procurement decision are missing the point entirely. What is required, in his view, is the refoundation of the operating model itself. The traditional structure of paralegals, juniors, seniors, and partners collaborating through well-defined steps is being fundamentally disrupted, and a huge part of that work will be automated or handled by AI agents supervised by practitioners acting as architects and controllers. "You, as practitioners, are those architects and controllers," he said.

For Leduc, firms that approach AI as a software selection exercise are asking the wrong question entirely. The real shift, in his view, demands a fundamental rethink of how patent work is organized: who does what, at which stage, and with what support.

‍

The Question Most Firms Are Not Asking

"We are so focused on the standard work today—running from deadline to deadline, from file to file, office action to draft—that we spend so little time on how we can actually make life easier for the users of the IP system,” Ernicke said. “By demonstrating what is really important IP, communicating IP more effectively internally and externally, or discovering risks more efficiently."

In his view, efficiency is the obvious first win from AI patent tools, but it is not the destination. The firms that will define the next decade of patent practice are the ones that use recovered time to ask a different question: what do our clients actually need, and are we delivering it?

Dr Hofmann offered a concrete example of what that shift looks like in patent prosecution practice. Today, writing an office action response is largely a mechanical move toward grant. Tomorrow, in his view, that same response could map to whether a dependent claim supports a standard-essential patent, or flag portfolio-level implications that currently go unnoticed because the manual workload leaves no room for them. 

"[It’s] something that at the moment many patent attorneys do not yet offer because it would just be too much work done manually."

‍

How Will AI Change the Billable Hour Model for Patent Firms?

Ernicke raised the topic of pricing with directness. 

In his assessment, patent firm pricing structures in Germany have been built around the same framework since the German Chamber of Patent Attorneys published its standard fee structure in the 1960s, he estimated. 

"Our whole pricing structure is still based on this old idea that comes from official fees and tries to adapt to this function, but was never actually value-based,” he continued. “It's a complete shift in paradigm of how we structure and price services…less focused on the procedural steps we have today."

When an office action response that took six hours takes two, Leduc observed, the billable hour model starts working against the firm. In his view, the transition to value-based pricing is not a distant theoretical concern, but a structural consequence of the efficiency gains already being delivered by AI patent drafting and prosecution platforms. 

"Firms will probably start thinking about what they sell," he said. "Maybe less hours, maybe more judgment, more added-value tasks. This is something that can start today, even if we don't need to get an answer right now."

‍

What the Combined Platform Delivers for German Patent Practice

The combined DeepIP and PatentMaker platform represents, in Leduc's words, the answer to what patent workflow automation looks like in a genuinely AI-native world: end-to-end coverage of the patent lifecycle, jurisdictional depth built from inside real practice, and the service layer required to make the transformation stick.

Wei, who has spent a decade working alongside German IP practitioners, framed it from the client side. "This teaming up with DeepIP will give us a chance to [build] broader roadmaps and more capabilities that you could have in your daily workflow," she said

For German firms specifically, Leduc described the acquisition as closing a gap that he sees as the defining challenge of the European market. Germany, in his words, is less transactional than other markets; practitioners are highly demanding, and earning their trust takes time. 

"Trust is everything, and trust can't be bought—you have to earn it," he said. That, he argued, is precisely what Dr Hofmann and Wei built at PatentMaker over the past years, and what DeepIP is now building on.

What the acquisition signals, ultimately, goes beyond the product. Every participant in the conversation returned, in different ways, to the same conclusion: the firms that thrive will be the ones that treat this moment as an organizational question, not a software question.

D'Archimbaud's closing remark said it best: "Don't treat it as a tool choice. Treat it as a transformation."

‍