Last month I wrote about why product decisions at Google felt different — the infrastructure, the synthesis layer, the people whose job was to connect scattered signals into a picture of what to build next.
The obvious follow-up: what does that look like when you're not Google?
Easier to show than explain. I recorded a quick video of our AI PM doing exactly this — connecting the signals, surfacing what matters (scroll to the end if you want to skip straight to it).
The problem: every feedback item needs context you don't have time to get
Every product team has a feedback channel. Support tickets, NPS surveys, reviews — it's all flowing in. You know it's valuable. You also know you're barely keeping up with it.
The reason isn't laziness or bad process. It's that every single item needs context before you can act on it.
A ticket says "onboarding crashed." Is it one user or systemic? Is this the first report or the twentieth? Is the feature broken for everyone or just this one CSV format?
Getting that context means checking analytics, looking up the user, searching for similar reports. 10 minutes per item. For 40 new items a week, nobody's doing that.
So the feedback sits there. And the insights stay buried inside it.
What surprised me while building Thrive
The thing I didn't expect was how much of a PM's time goes to confirming things AREN'T fires.
Here's an example from the demo. A ticket comes in: "exports not working." Sounds like it could be bad. Could be a regression affecting thousands of users.
But when you check: 1 user out of 97,000. Specific CSV format, edge case. No other reports.
Not a fire. Move on.
That investigation took 10 seconds. The manual version takes 10 minutes. And for every ticket that IS a fire, there are probably 10 that aren't. Knowing the difference quickly is what lets you focus on the things that actually matter.
Every piece of feedback — tickets, surveys, reviews — contains more information than what's written in it. The question is whether you have time to extract that context for each item. (Spoiler: you don't. Nobody does.)
What happens when you have an AI PM on your team
I recorded a 60-second walkthrough of what this looks like end-to-end — how every item in your feedback channel gets context automatically, how individual tickets get investigated in seconds, and what happens when you just ask "what's been going on with my users this week?"
Try it out with your own feedback data! I’d love to know your thoughts.
— Ishwar, ThriveAI