Consumer-facing features arrive with legal questions attached: consent, age gates, marketing claims, data handling. Product counsel answers them, and the answers are due inside the sprint, for each jurisdiction the feature ships to.
That pressure is not new. What is changing is the form the answer takes. AI for product counsel, as the lawyers in this piece use it, means research that comes back with cited primary law, and legal judgment that leaves the room as something engineering can build from.
One general counsel showed what that looks like in practice.
The night Rachel Harris decided to bypass the launch-review handoff, she built something instead. Harris, GC and AI Governance and Privacy Officer at Suzy, wrote a legal spec for a missing consent feature and took it straight to her head of product and head of engineering. She told the story on CZ and Friends, the podcast where GC AI CEO Cecilia Ziniti interviews legal leaders:
"What this new technology allowed me to do last night was... go back and forth, get the concept down, create an MD file of here is how it would conceptually look from the legal perspective... in these jurisdictions, consent might look a little different. You're gonna need to understand if it's under a certain age, is it in, you know, Europe versus the US... They were able to immediately download the MD file, plop it into their agents, and say, start building me a prototype. And that's where we are this morning."
By the next morning, engineering was prototyping from it. Her CEO put a name on the pattern:
"Can you infuse the brain of Rachel into the thing that you've built?"
Her judgment left the room as a document their agents could build from, and the launch-review meeting became a launch-design meeting.
Harris's build shows where the role is headed.
Product counsel is the in-house role AI is redefining first, because the job already runs at engineering speed: consumer-facing features, time-sensitive advice, and a regulatory map that moves mid-sprint.
GC AI is the enterprise legal AI platform a three-time general counsel (Anki, Bloomtech, and Replit) built for in-house teams. Five GC AI features do the product-counsel work in this piece:
Research answers the jurisdiction questions with cited primary law, US case law included.
Projects holds the context of a launch across each chat about it.
Custom Company Profile (GC AI's team-level standards setting) keeps outputs calibrated to how your company takes risk.
Exact Quote keeps citations verifiable, character by character.
GC AI's API carries that same context into the tools engineering already uses.
What Product Counsel Does Now With AI
Product counsel sits with product and engineering teams and clears the legal path for what they build. The work is consumer-facing, time-sensitive, and jurisdictional. Teams rely on you, and they do not want to wait for legal.
Three pressures land on this role at once, where other in-house work spreads them out:
Advice that ships to users at scale, so a wrong call reaches them at scale too.
Timelines set by engineering sprints.
A regulatory map that moves mid-build, jurisdiction by jurisdiction.
That is why strong product counsel tend to sit closer to their engineers than to the rest of the legal team.
Bill Berry, General Counsel at Verkada, described the structural problem: the role means "trying to fit yourself in a sequential process. That's not really sequential from a legal side."
Engineering runs in sequence. Legal questions do not. They come in from several directions at once, and the product counsel sorts them into two piles: the ones that stop a launch, and the ones that can ship with a risk the company has agreed to carry.
Molly Abraham, General Counsel at Coinbase, describes the other half of the job, keeping the right people in the loop:
"One of the most important roles that we serve as product counsels... is to be the dot connector across not only the legal organization, but policy, and compliance, and the business... We work in channels. We don't work in DMs on Slack."
Why AI Redefines the Product Counsel First
Much of product counsel work is the same question asked again across jurisdictions, on a deadline. That is the kind of work AI handles well. A consent-flow question that spans the EU, the US, and an under-16 audience used to mean a research memo or a call to outside counsel. With AI research that cites its sources, it becomes an afternoon of checking the citations. Chat 2.0, GC AI's multi-agent research and drafting engine, shows that step on screen:
The bigger shift is the output. Advice that used to arrive as comments on someone else's spec now arrives as the spec itself, in the same formats engineering already works from. The prototype conversation starts from the legal design instead of after it. Stanford Law School's product counseling podcast framed the same shift in July 2026: legal teams are advising alongside the AI, and the role is being rebuilt around that fact.
Abraham frames the mandate as three hats for legal: AI enabler, AI protector, and AI super user.
"I don't think law is going to become extinct. I don't think in-house legal is going to become extinct. I think in-house lawyers who do not embrace AI will become extinct."
From Reviewing Launches to Writing the Legal Spec
A legal spec is a short document, not a memo. It states the legal requirements in the form engineering already uses to build features. For a consent feature, the skeleton looks like this:
Feature: what ships, and where
Jurisdictions in scope: EU, US federal, US states with age-verification laws
Consent standard per jurisdiction: opt-in vs. opt-out, parental-consent thresholds
Age thresholds: what changes under 13, under 16
Fallback positions: what ships if the preferred design slips
Decision rights: what engineering can ship without legal, what needs sign-off
Open questions for engineering: data retention, geolocation, verification UX
Citations: the primary law behind each position
The working pattern from Harris's build, generalized into steps a product counsel can run this sprint:
Pick the recurring launch question: the consent flow, the age gate, the claims review that arrives with each feature.
Write the legal spec once: the jurisdictions, the thresholds, the fallback positions, in a document engineering can drop into its own tools.
Ground it in cited research: run the jurisdiction questions through Research and keep the primary-law citations in the spec.
Keep it living: hold the spec and each follow-up in a Project, so the next feature starts from the last answer.
Hand it to the builders: the spec goes to product and engineering as an input, and review time drops to verifying what came back.
Each spec compounds. The second launch starts from the positions the first one settled, and over time the product team is building on the product counsel's judgment without booking a meeting for each question.
Product Counsel vs. Generalist In-House Counsel
A generalist in-house counsel covers the company's whole legal surface, so product questions wait in the same queue as everything else. Product counsel is what that slice of the work becomes once launches move fast enough to justify a dedicated owner: the same duties, a narrower surface, and a faster clock.
If you're deciding when to split the role, our corporate legal department structure guide walks through it. The signal to watch is the queue: if product questions keep waiting behind contract work, your launches are absorbing the delay.
Where GC AI Fits the Product Counsel Job
GC AI was almost named Product Counsel AI. Today it is the legal AI platform for 2,000+ in-house teams in 53 countries (Sep 2026), including 200+ public companies and names like SKIMS, Duolingo, and Bass Pro Shops. As Tiffany Lee, General Counsel of Liquid Death, said in Bloomberg Law: "For in-house legal teams, AI is the tech unlock that has actually delivered massive ROI."
For product counsel, that unlock runs as one workflow: the research, the spec, the handoff. You can see it end to end in our demo:
The API Handoff
The handoff Harris did by hand, spec to engineering, can also run through GC AI directly. GC AI's API lets product and engineering teams put the same questions to GC AI from inside their own tools, with the company profile and the lawyer's positions already attached.
Giving product teams legal capability while the lawyer keeps the final call is the architecture Cecilia Ziniti has argued for. The non-lawyer side of that handoff is already real. About a third of the people on the GC AI platform aren't lawyers at all. We have contract managers and procurement professionals using GC AI to make their work easier, more accurate, and faster.
The Privilege Question
Before you compare features, ask about privilege. Coinbase's Abraham said:
"I definitely think folks should be thoughtful about privilege. Because a non-lawyer asking an LLM for legal advice is not necessarily their lawyer."
That's why this work belongs in an enterprise legal platform, not a public chatbot. GC AI works with leading AI providers such as OpenAI and Anthropic, none of which train on your data, and maintains zero-data-retention agreements with its LLM providers wherever feasible. The full subprocessor list is public.
Where to Go Next
New to legal AI? Start with GC AI's legal AI classes, taught by former general counsels. More than 6,000 lawyers have completed them.
Want to read more first? AI for general counsel operations covers the department-wide view, and in-house counsel AI software compares platforms for a whole team. And when you're ready to run this on a real launch, start with the free trial below.
Write the legal spec once, and each launch after it starts from legal's answer.
Frequently Asked Questions
What Is the Difference Between Product Counsel and Commercial Counsel?
Product counsel advises on what a company builds: features, consent flows, claims, and data handling inside the product, working directly with product and engineering teams. Commercial counsel advises on what the company sells and signs: customer contracts, vendor agreements, and negotiations. The two roles overlap at launch, when the feature meets the order form.
Can AI Help Product Counsel Review Feature Launches Faster?
Yes, in two ways: cited multi-jurisdiction research turns launch questions from memo-length into same-afternoon answers, and a written legal spec lets engineering build from legal's positions instead of waiting on review. One working product counsel drafted a consent-feature spec with AI and had engineering prototyping from it by the next morning.
Do Companies Need a Dedicated Product Counsel?
A dedicated product counsel makes sense once launch velocity outruns the general legal queue. Before that point, a generalist in-house counsel covers the product-facing work, and AI leverage extends how long one lawyer can hold that slice.
