Notes from the IP Counsel Cafe Silicon Valley Meeting on what AI is actually doing for in-house IP teams: real efficiency, fragmented workflows, and the growing gap between client expectations and firm reality.
Field notes from the IP Counsel Cafe Annual Silicon Valley Meeting, 2026. Conference theme: "Value Under Pressure: IP Strategy When Scrutiny is High, Budgets are Tight, and AI is Advancing."
This week, I attended the IP Counsel Cafe Annual Silicon Valley Meeting in Mountain View, which is a great gathering of the IP community. The theme this year was "Value Under Pressure: IP Strategy When Scrutiny is High, Budgets are Tight, and AI is Advancing." A few things stuck with me. I have compiled them here, less as a comprehensive recap and more as the moments that are still rattling around in my head.
IP teams are navigating a perfect storm. Budgets are tighter, expectations from business leadership are higher, and AI is transforming patent work faster than most organizations can adapt. Across nearly every session, the same three dominant themes kept surfacing: AI adoption in IP departments is real but uneven, IP teams now have to prove business value and not just legal value, and the USPTO and litigation landscape are shifting in ways that change prosecution strategy.
What follows is eight takeaways grouped under those three themes, written from the perspective of someone trying to make sense of where in-house IP work is actually heading.
Most in-house IP departments are now using AI somewhere in the patent lifecycle, but adoption is fragmented and individual-led rather than end-to-end. The most common use cases are invention capture, prior art search, drafting provisionals, and office action response preparation, where teams are seeing genuine efficiency gains. Higher-stakes work, including strategic decisions, complex §103 responses, and legal opinions, still requires deep human judgment. There is also a notable disconnect between what in-house counsel expects AI to save on outside counsel fees and what those firms are actually seeing in efficiencies.
Almost every team is using AI somewhere in the patent process, but "using AI" today means individuals running pilots with approved tools, chipping away at discrete tasks, and sharing what works internally. It does not mean a coherent, end-to-end workflow. The vision of agents gracefully strung together from invention disclosure through prosecution and portfolio management is not a reality yet. What is actually happening is productive but fragmented: one person finds a great use for prior art search, another automates office action prep summaries, someone else is experimenting with invention harvesting from Slack and email. That is real progress, but it may not qualify yet as transformation. This is one of the main reasons measuring AI ROI in IP departments is so challenging at this stage.
The barrier to AI adoption is not willingness or budget. It is more often capacity and access. Most teams want to experiment. The problem is that lean IP departments do not have excess time to evaluate tools, build agents, redesign workflows, and train colleagues on top of their existing workload. Getting new tools through infosec and procurement can take months, which means many teams are effectively forced to innovate within the constraints of whatever enterprise-approved software they already have. That can still yield results, but there is no way to simply trial anything and everything available and narrow it down from there.
A third bottleneck is starting to emerge: the people who push AI forward in their organizations tend to be technically fluent and genuinely curious about building, and there simply are not enough of them. Organizations that are pulling ahead are starting to carve out dedicated roles or identify internal champions who have both the legal knowledge and the technical aptitude to bridge the gap.
"The barrier to AI adoption is not willingness or budget. It is more often capacity and access."
Management is reluctant to add headcount because the prevailing assumption is that AI should be able to absorb the load. But the teams on the ground who would actually build and deploy those AI workflows are already stretched thin and do not have the bandwidth to do the upfront investment required to create the efficiencies management is expecting. Someone has to put in the effort of making AI work, and right now, that work is often falling on people who have no room for it. Until organizations resolve this tension, either by protecting time, adding dedicated AI-capable roles, or resetting expectations, the efficiency gains will remain theoretical for many teams.
The honest read from practitioners on the ground is that AI is making patent work better, but it is not yet meaningfully compressing the hours. AI is improving quality more than it is saving time. The time savings that justified the investment pitch to management may still be 12 months or more away, as workflows mature, tools improve, and teams build the muscle memory to use them fluently. Setting realistic expectations internally about the near-term ROI profile (quality first, speed later) will help avoid disillusionment.
"AI is improving quality more than it is saving time."
In-house teams broadly want their firms using AI, and that is a notable shift from a year ago when many clients were still saying no. But the expectation that AI translates into lower fees has not fully materialized yet, according to the firms. Most firms are still experimenting, while some clients are mandating flat fee cuts on prep and prosecution. Transparency about which tools are being used and when is becoming a baseline expectation, not a differentiator. In-house teams expect their firms to show efficiencies that are achieved safely and quickly while also keeping work product quality high. That is a lot of pressure for firms, but the firms that can deliver on it will ultimately win.
Want to see how Juristat fits into an AI-augmented IP workflow?
Explore Juristat for IP teamsBusiness leaders increasingly see IP as a strategic asset, which raises the bar for what IP departments need to deliver. Compliance checkboxes are no longer enough. Legal teams are being pushed to produce patent valuations, competitive intelligence, revenue generation strategies, and demonstrable ROI. Companies that actively monetize IP are no longer treated as cost centers. Those that cannot quantify the value their IP generates often still are.
The "cost center" framing is a strategic liability for IP departments. Winning teams are quantifying what patents deter, what counterfeits they have taken down, what market share they have protected, and what licensing or partnership revenue IP has enabled. The portfolio needs to be understandable to the CFO and the GC, and value needs to be conveyed via KPIs that matter to the business, not metrics that only make sense inside the IP team.
Several USPTO and litigation trends are changing how IP departments think about prosecution strategy. Appeal pendency has improved significantly, making appeals (including on §101) a more viable strategy alongside or instead of RCEs. Fee increases are putting continuation strategy under the microscope. IPR and discretionary denial trends are affecting litigation strategy, and the re-examination landscape is evolving. Through it all, examiner quality concerns persist as examiners are pushed to meet their goals.
With appeal decision pendency down roughly 50% and PTAB showing more receptivity on §101 than examiners, appeals deserve a place earlier in prosecution strategy rather than as a last resort after an RCE. The economics are also shifting. With fee increases putting continuation strategy under pressure, the cost-benefit comparison between a continuation and an appeal looks different than it did even two years ago.
"Appeal decision pendency is down roughly 50%, and PTAB is showing more receptivity on §101 than examiners."
There is a perception that with the rise of AI, in-house teams will be able to take on more and outsource less to firms. That may be a trend over a longer time horizon, but the general consensus at the conference was that this is not happening widely yet, and that wholesale insourcing would actually prevent in-house attorneys from focusing on strategic work. The smart move right now is being more deliberate about what gets sent out and to whom, rather than insourcing en masse. For many teams, the near-term opportunity is to be more selective about outside counsel and to hold them to a higher bar on quality, speed, and AI-informed efficiency.
As always, the best part of IP Counsel Cafe is having a chance to connect on a personal level and really reflect. There is something special about a room full of practitioners who are wrestling with the same hard problems and having honest conversations about what is working and what is not.
Get more analysis on AI adoption and IP department strategy delivered when we publish.
Strategize With JuristatMost in-house IP departments are using AI for discrete patent prosecution tasks rather than end-to-end workflows. The most common use cases are invention capture, prior art search, drafting provisionals, and office action response preparation. Adoption is fragmented and typically driven by individual attorneys experimenting with approved tools, not by coordinated department-wide deployments.
The biggest barriers are capacity, infosec and procurement timelines, and a shortage of attorneys who are both legally fluent and technically capable of building or deploying AI workflows. Budget and willingness are rarely the bottlenecks. Lean IP teams do not have the time to evaluate tools and redesign workflows on top of their existing workload, and getting new tools approved can take months.
Not yet in a measurable way at most organizations. Practitioners report that AI is improving the quality of patent work more than it is compressing hours. Meaningful time savings are likely 12 months or more away as tools mature and teams build proficiency. Setting near-term expectations around quality improvement first, time savings second, helps avoid internal disillusionment.
Not consistently. In-house teams broadly expect their outside counsel to use AI, and transparency about tool use is becoming a baseline expectation. But most firms are still experimenting and have not translated AI use into significantly lower fees. Some clients are mandating flat fee cuts on prep and prosecution to force the issue.
Yes, significantly. Appeal decision pendency is down roughly 50%, which makes appeals (especially on §101 rejections, where PTAB is showing more receptivity than examiners) a more viable strategy earlier in prosecution rather than only as a last resort after an RCE.
Not wholesale. The current consensus is that AI may shift the in-house and outside counsel balance over a longer time horizon, but aggressive insourcing today would pull in-house attorneys away from strategic work that only they can do. The near-term opportunity is being more selective about what gets outsourced and holding outside counsel to a higher bar on quality, speed, and AI-informed efficiency.