Most IP teams have moved past the question of whether to use AI. The questions that matter now are harder: Where does AI belong in the patent lifecycle? How do you trust what it produces? And who is accountable when it isn’t perfect?
Watch the full webinar below:
Those were the questions behind “Idea to IP: Turning Inventions into Patents with AI,” a ChIPs Network accelerator hosted by Solve Intelligence. The session was moderated by Bree Vculek, a patent attorney and Legal and Product Engineer at Solve.
It brought together four outstanding practitioners who cover the full patent lifecycle, covering both in-house and outside counsel perspectives:
Here’s what we discussed:
When asked what works differently now than it did two years ago, Dina answered simply: “The entire process is different,” from how inventions are harvested to how applications are written and prosecuted.
Larissa shared a story many attorneys will recognize. A client was about to disclose an invention in an acquisition meeting, and the filing instructions got stuck in his outbox. Her team didn’t receive them until 1 p.m. on the day the application had to be filed. “We were able to turn it and do this beautiful patent application, and the client was like, ‘How did you do that?’” Before AI, she said, a same-day filing meant changing a few things, adding a cover sheet and submitting.
Kate described a similar shift on the in-house side. Prior art searching and product charting “would typically take weeks or months at a time, whereas now we can turn that around in a few hours to a few days.” At Verizon, Vaishali’s team is also using AI to analyze which patents to maintain and pay maintenance fees on, much faster than before.
Each panelist described becoming an AI leader in her organization in a different way. Kate kept volunteering for AI-related work at her firm and became the go-to AI contact for her lean in-house team after moving to BlackBerry. Larissa’s background representing clients on AI inventions made her a natural spokesperson, and in her own words, “[i]f we did not embrace and fully change how we operate, we were going to be left behind.” Dina started working with AI in 2018 when a client’s AI-heavy work landed on her desk, and she later moved into AI governance and tool evaluation for the firm.
When evaluating tools, the panel’s priorities were consistent:
Dina noted that clients have gone from asking “Are you using AI?” to “Why aren’t you using AI?” They now not only expect a clean product free of typos, but also “a superior work product with more forethought and strategy…just because we have more information available at our fingertips.”
Kate doesn’t wait for engineers to come to her. She looks through internal systems, joins team meetings and quarterly reviews to spot new ideas early. Then she reaches out directly: “I saw this cool feature that you mentioned… I’d love to hear more about it.”
AI speeds up each step. Kate uses Solve to turn code and documentation into flowcharts that help her understand an invention quickly. She also runs semantic prior art searches instead of keyword strings, which has cut her review time “from a few days to sometimes a few hours.”
Further, at Verizon, deciding what to file means weighing business strategy, standards contributions, product use, detectability, industry adoption and patentability. Vaishali said AI helps her team map the landscape and decide where to spend resources. It has also helped keep outside counsel’s drafting costs steady while budgets fluctuate.
Larissa called rules and constraints “probably the thing that AI does the best.” Every partner has a style, and every client has a sub-style. Her team builds client-specific templates in Solve that cover best practices.
Those templates solve a very human problem. Associates working with partners and clients “forget which partner wants to say ‘can,’ which partner wants to say ‘may.’” Now, she said, “there’s a gatekeeper that keeps everything confined such that we know it’s going to be delivered in the right way.”
What happens to the time AI saves? Larissa said that for fixed-fee clients, her firm can now reliably deliver within the agreed budget, and is “effectively writing off less time.” For higher-value matters, such as life sciences applications, the savings can fund additional services.
When one semiconductor client asked her firm to cut fees by a quarter, Larissa agreed, on the condition that her team could use AI. Her firm rolled out Solve, and the client came back saying, “This is the best patent I’ve ever read.”
Her firm also ran a “stopwatch test.” Four applications were drafted by hand and then re-drafted from the same material using Solve. The AI-assisted versions saved time, and Larissa said “it was absolutely wild how similar the applications were.” Her explanation was that a patent specification is a structured, repetitive kind of writing, which AI does very well “given the right guidance and bumpers to stay within.”
Dina always checks whether an office action holds up before responding, and she uses Solve to do it. “Solve has this tool where they’ll just do a claim chart of all the limitations and all the rejections. So it’s very easy to see which art applies, which one doesn’t, and where the examiner was vague.”
She said Solve does one thing particularly well: “They give you options.” Solve suggests several possible amendments on different limitations, and she can redirect the AI to focus on a particular limitation instead. Solve also shows exactly where in the references the art is cited, which saves hours of review time.
Vaishali, drawing on her time as Commissioner for Patents, gave a direct warning: not everyone is using AI well. She recalled examiners’ frustration with filings that were clearly AI-generated and unchecked. Examiners face heavy workloads and deadlines, and well-prepared applications and responses make their jobs easier.
Larissa agreed, describing drafts with invented claim limitations or 70 pages of unnecessary text. Her rule for her team is simple: “You cannot send me something that you haven’t actually read. There has to be a human in the loop.” Dina reminded associates that “just because AI wrote it doesn’t mean it’s correct.”
Dina said collaboration depends on how comfortable the client is with AI. Sophisticated AI companies sometimes hand over nearly complete applications, which her team refines and strengthens in Solve. Other clients still start with a traditional invention disclosure form.
From the in-house side, Kate said she can now give outside counsel a “more focused package.” She runs her own prior art searches, which can return results in minutes when a feature changes, and then the direction can be set by the patent committee well before the first call. That means fewer emails, fewer phone calls, and a clearer claim strategy from the start.
Kate uses AI in almost every licensing task. She uses it to check targets against business criteria by searching public documentation, analyzing YouTube transcripts and figures for evidence before deciding to download gigabytes of data. “Sometimes if you’re spending hours or days at a time, you kind of convince yourself that this mapping makes sense.” AI gives her a second set of eyes before peer review.
Vaishali described how her team used AI to prepare a corporate representative for trial. They summarized deposition transcripts so he could prepare efficiently. They also built witness outlines in the style of the opposing counsel's past questioning. The corporate representative “was blown away, he said it made him feel so prepared, and he was just a fantastic witness.”
Dina said well-built templates do much of the work, because prompts and constraints are built in. At some point, though, firm training needs to be specific to each tool.
Larissa mentioned a law professor’s estimate that it takes about 100 hours of working with AI to use it well. Realizing she needed more practice herself, she now meets with Solve every month to learn a new way to use the tool effectively in her practice. “If I’m going to make other people use AI, I guess I should be using it too.”
For teams starting to implement AI, the panel offered practical advice. Dina said IT and innovation teams will become “your best friend” for setup, template creation, and customized training. Larissa relies on fast, responsive support from Solve. Vaishali recommended asking peers what they use and taking advantage of trial periods. Kate emphasized involving security early, defining the specific tasks you want a tool to do, and testing it with your real matters before committing.
One theme ran through the entire conversation. The teams getting the most out of AI aren’t necessarily the ones using it the most. They’re the ones who have thought hardest about where AI is used and where the human review time is spent.
That idea shapes how we build Solve. It’s why our office action analysis tool charts each limitation against the cited art, why drafting templates capture attorney preferences alongside client instructions, and why we work closely with 700+ IP teams to keep improving.
Thank you to Dina, Larissa, Vaishali, and Kate for such a candid conversation, to the ChIPs Network for partnering with us, and to everyone who joined and asked great questions. We look forward to seeing many of you at the ChIPs Global Summit later this month.
If you’d like to pilot any of the workflows discussed, get in touch with Solve Intelligence.
