This fall, in-house IP teams are facing a familiar ritual: pull last year's numbers, brace for a budget cut, and figure out what to trim without gutting the portfolio's real value. On Tuesday, I participated in a IPWatchdog webinar with Gene Quinn and Jason Harrier (AI Patent Counsel, Salesforce) that tackled this exact problem: how AI is reshaping the budgeting and cost-cutting process for 2027, and what teams of any size can do about it right now.

The rules for challenging a competitor's patent have fundamentally changed in 2026. Inter partes review (IPR) petitions have dropped 65% year-over-year to historic lows. Ex parte reexamination has overtaken IPR as the leading post-grant validity tool for the first time since the America Invents Act. Post-grant review (PGR) windows are closing fast, and third-party pre-issuance submissions are drawing renewed attention from IP strategists. If your patent challenge playbook hasn't been updated, it's already outdated.

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.

These five AI prompts help patent professionals make better prosecution decisions, forecast client outcomes, evaluate examiner interviews, protect claim scope, and identify business development opportunities using prosecution data.

Patent firms rarely lose margin because demand disappears. They lose it when prosecution work gets harder to deliver at scale. The firms getting this right are tightening intake, pricing around real prosecution complexity, automating office action response and IDS work, and winning more business from the clients they already serve.

Better patent data leads to better patent strategy because patent AI is only as useful as the patent-specific data behind it. When tools are grounded in a strong data foundation, they can produce better strategic outputs, stronger patent workflow support, and better patent decisions.

Patent practice depends on nuance, legal reasoning, client goals, and risk assessment. The most trustworthy AI in patent practice supports attorneys with better patent-specific data and stronger workflow fit, helping them make better strategic decisions without replacing professional judgment.

Generic AI is not enough for patent work. Patent-specific data makes AI more trustworthy, more useful in prosecution workflows, and easier for practitioners to defend to clients and partners.

The USPTO is not the same office it was two years ago. Since January 2025, shifts in leadership, policy emphasis, and internal guidance have changed how applications are examined and how prosecution outcomes are determined. Some of these changes have generated significant attention in the patent community, while others have been more subtle. All of them have real consequences for practitioners and applicants trying to build and protect patent portfolios today.

AI can help patent teams reduce manual work and improve workflow efficiency. The challenge is adopting it in a way that supports attorneys, maintains quality, and aligns with real prosecution practice.