AMA: Figma AI Product Leader, Rodrigo Davies on AI Product Management

A few key principles I always try to follow:1. Transparency - it should always be obvious who is taking an action (a person or AI). Clarity trumps "magic" every time.2. Auditability - there should always be a record of every step AI takes, in case the user or product team needs to debug later. The more of that you can show to the user in a useful form, the better. Enabling users to self-diagnose and solve issues is empowering and trust building.3. Expect to need off ramps and controls - agents w ...Read More

For me the gold standard output metric is feature retention. If users continue to want to use agentic AI, that's the best proxy for whether it's useful. However, there are clearly other workflow metrics that are specific to the product space you're operating in. For example, if agentic AI is supposed to help customers do something they've always done, but more efficiently, you should see significant increases in those local metrics. Qualitative signals like thumbs up/down and in-product surveys ...Read More

I would encourage play and experimentation as much as possible. A great place to start is using AI tools with side projects and things that folks on the team are super interested in - even if they're not work related. That'll demystify the technology and help folks get a sense for what's possible, what's hard and what's easy. To get that started, I've found that showing real examples of AI tools in use is powerful. The internet is full of splashy 30 second videos of the "happy path" of using AI, ...Read More