Introduction
Stilta, a company developing an AI platform for patent litigation and IP intelligence, has announced that it raised $10.5 million in a seed round led by Andreessen Horowitz, commonly known as a16z. In addition to Y Combinator, the round included founders and operators associated with companies such as Sana, Legora, OpenAI, Lovable, and Listen Labs.
Stilta aims to use AI to automate the analysis of the vast patent portfolios held by companies and law firms, and to apply that analysis to patent enforcement, licensing, competitive analysis, and other purposes. According to the company’s announcement, its AI agents reason across more than 180 million patents and output their findings in a form that can be traced back to specific documents.
What makes this news interesting is that it is not simply another example of AI entering patent work. Rather, it seems to indicate that the meaning of patent portfolios is beginning to change for patent-owning companies, patent firms, law firms, and even competitors.
AI That Unearths Dormant Patents
Many companies hold large numbers of patents as a result of their research and development activities. However, not all of those patents are necessarily being fully utilized in business.
A patent does not become valuable in practice simply because it has been obtained. It begins to demonstrate practical value only when its relationship to other companies’ products is examined, its scope of rights is assessed, invalidation risks are estimated, and materials capable of supporting negotiations or litigation are organized.
This work is extremely labor-intensive. It requires reading patent claims, reviewing specifications, investigating prior art, and comparing the patent with the specifications of the target products or services. Moreover, when many patents or products are involved, it is not realistic to conduct a comprehensive review by human effort alone.
As a result, some companies own patents that may have value but do not fully understand or utilize that value. This “underutilized area of intellectual property” is precisely what Stilta appears to be targeting. Its distinctive feature lies in using AI to explore relationships between patents and markets, products, and competing technologies that were previously too extensive to review manually, thereby identifying candidates for enforcement or licensing.
The Possibility of Lowering the Barrier to Enforcement
Oskar Block, CEO of Stilta, has made a statement to the effect that once one company begins using AI for patent enforcement, other companies in the same industry will be forced to choose whether to do the same, pay license fees, or expose themselves to risk.
Although this is a somewhat strong statement, it succinctly captures the changes taking place in IP practice. Traditionally, patent enforcement has involved significant costs. This is because many steps are required, including selecting target patents, examining the possibility of infringement, collecting evidence, conducting invalidity searches, and obtaining expert reviews.
However, if AI lowers the cost of initial analysis, patents that previously would not even have been considered may become candidates for licensing negotiations or warning letters. For patent holders, this means an expansion of opportunities to use their rights. On the other hand, for companies that provide products or services, it means an increased need to detect third-party patent risks at an earlier stage and take defensive measures.
In other words, rather than automating patent litigation itself, AI may significantly enhance the exploratory capabilities that precede patent litigation and licensing negotiations.
What Matters Is Not “the AI’s Answer,” but the Ability to Return to the Grounds
One of the notable points in Stilta’s announcement is that the company says findings generated by AI agents will be output in a form that can be traced back to specific documents.
In IP practice, it is not enough for AI simply to produce conclusions that appear plausible. Which element of which claim corresponds to which statement in which product information? Which prior art document relates to which feature of the claimed invention? Based on what evidence was a particular inference made? Unless practitioners can return to these grounds, the output is difficult to use as a basis for practical judgment.
In the world of patents, what ultimately matters is language and evidence. It is necessary to carefully connect claim language, descriptions in the specification, prosecution history, prior art documents, product materials, technical explanations, and other materials. For this reason, in IP AI, not only summarization and search capabilities but also the ability to present grounds, explain correspondences, and ensure verifiability become important.
In this respect, Stilta’s direction can be seen as being close to the essence of AI utilization in the IP field.
How Will the Roles of Patent Firms and Law Firms Change?
As AI platforms of this kind become more widespread, the roles of patent firms and law firms are also likely to change.
First, parts of investigation and initial analysis may be made more efficient by AI. Areas such as taking inventory of patent portfolios, extracting candidates for enforcement, generating hypotheses about correspondences with competing products, and searching for potential invalidity materials are well suited to AI.
However, this does not mean that the role of experts will disappear. Rather, as AI begins presenting large numbers of candidates, human experts will increasingly be required to evaluate those candidates, organize them into legally meaningful arguments, and assess risks.
The interpretation of patent claims, the possibility of infringement under the doctrine of equivalents, the relative strength of invalidity grounds, litigation strategy, how to present arguments in negotiations, and possible settlements based on the other party’s business structure cannot be resolved by document search alone. AI may expand the domain of “finding,” while increasing the importance of the human domain of “judgment.”
Defensive IP Intelligence Will Also Become Important
Stilta’s news tends to be discussed as a tool for enforcement, but it is also important for the defensive side.
If other companies begin using AI to explore correspondences between their patents and a company’s products, that company will likewise need to identify risks at an early stage. It will become necessary to examine relevant patent risks more broadly before launching new products, when changing specifications, when considering M&A or partnerships, and when expanding overseas.
In technical fields where enormous numbers of patents exist, responding only after a problem has surfaced may be too late. AI-based IP intelligence may be used not only for offensive licensing strategies, but also for defensive clearance, design-around efforts, invalidity searches, and preparation for negotiations.
In this sense, the spread of IP AI will not merely strengthen rights holders. Rather, it can be understood as raising the speed of information gathering and analysis concerning patent risks across entire industries.
The Gap Between “Usable Patents” and “Unusable Patents” Will Widen
Another important point is that as AI becomes more widespread, the quality of patents will be scrutinized more rigorously.
As AI makes it easier to analyze entire patent portfolios, the difference between patents that merely increase the total number of rights and patents that are actually useful for enforcement or negotiation will become more visible. It will become important once again to consider whether the claims can readily read on target products, whether the specification describes sufficient embodiments and variations, and whether the patent covers technology areas that are important to the business.
This will also affect practice at the filing stage. If patents are expected to be explored, compared, and evaluated by AI in the future, it will not be sufficient simply to create patents that can be granted. It will become increasingly important to design patents that have a clear relationship to the business, can withstand comparison with other companies’ products, and can be easily explained in the context of licensing or litigation.
In the era of IP AI, the issue will not only be the “quantity” of patents, but also the “quality” that allows their value to emerge when analyzed.
Implications for Japanese Companies
This development is also relevant to Japanese companies. Japanese companies have accumulated many patents through years of research and development, but they are not necessarily using all of them strategically.
Going forward, it will be important not merely to manage patent portfolios as assets to be held, but to analyze them in connection with business, competitors, standardization, licensing, M&A, and litigation risks. In that process, AI can serve as a powerful aid.
However, introducing an AI tool does not immediately elevate an IP strategy. Without strategic axes—such as which technology areas to prioritize, in which markets to use rights, which competitors to focus on, and how much risk to accept—it will be difficult to make effective use of the large volume of information presented by AI.
AI does not replace IP strategy. It expands the analytical capabilities needed to execute IP strategy. It is important not to misunderstand this point.
Conclusion
Stilta’s fundraising seems to indicate that IP AI is moving beyond a mere research support tool and toward becoming infrastructure related to the utilization of patent portfolios, enforcement, licensing, and defensive strategy.
A patent does not create value merely by being obtained. It becomes a business weapon only when it is understood which patent has what meaning, in which market, against which party, and in what context.
AI is beginning to significantly change the entry point for making that determination. In future IP practice, the question will not only be whether to use AI, but also how to evaluate the information discovered by AI and how to connect it to strategy.
Stilta’s news may be seen as a symbolic event in the shift of patents from “something to own” to “something to explore and use.”