Screenpipe (YC S26, $25/month, roughly 20,000 GitHub stars) gives founders a searchable local recording of everything they saw and said, genuinely useful for settling factual disputes. But Daniel Kahneman's WYSIATI research in Thinking, Fast and Slow shows more raw memory doesn't reduce overconfidence; only structured tools like the premortem catch the story-coherence trap that drives bad decisions.
I installed Screenpipe on my main machine the week it launched on Hacker News, 85 points, YC S26 batch, positioned as the open, local-first answer to Rewind and the now-discontinued Limitless pendant (Meta acquired Limitless in December 2025 and stopped selling the pendant to new customers). The pitch is seductive for anyone who runs a company: it captures your screen and audio 24/7, transcribes locally with Whisper, and lets you or an AI agent ask what you actually decided in the call on Tuesday. It's free to start, $25/month for the signed desktop build if you want it running without constant macOS security prompts, and it's already sitting at roughly 20,000 GitHub stars.
After two weeks of actual use, my answer to "does this make me a better decision-maker" is: it makes me a better rememberer, which is not the same thing, and conflating the two is exactly the trap Daniel Kahneman spent Thinking, Fast and Slow warning about.
Kahneman's WYSIATI principle, "what you see is all there is," says our confidence in a judgment depends on how coherent the story we can tell is, not on how much or how good the underlying evidence is. Give System 1 a plausible narrative and it stops asking what's missing. That's the risk with a tool that hands you an effortlessly searchable transcript of everything you said in a negotiation: it doesn't make the story more true, it just makes it feel more complete, which is precisely the cognitive-ease trap that produces overconfidence. More raw material for your remembering self isn't the same as more rigor in your reasoning self.
Where Screenpipe genuinely earns its place is in the boring, factual disputes founders actually have, like whether the vendor agreed to that price on the call, or what I actually told the team about the roadmap in June. That's memory doing memory's job. It's a materially better tool for that than my own recollection or a co-founder's notes, and the local-first, no-cloud design means I didn't have to trade privacy for it.
Where it fails is the harder job: making the decision itself better. For that, Kahneman's own answer is boringly effective and Screenpipe doesn't replace it: the premortem. Before a big call, imagine it's a year later and the decision failed, then write down why. That single ten-minute exercise catches the deception WYSIATI creates, because it forces you to argue against the coherent story you already believe.
Here's how I actually score the two approaches against what Kahneman's research says decision quality depends on:
| Criterion (Kahneman) | Screenpipe (memory) | Premortem (protocol) |
|---|---|---|
| Reduces WYSIATI overconfidence | No, more material, same bias | Yes, directly targets it |
| Forces the outside view | No | Yes, when done with base rates |
| Catches missing information | Partially, recalls what happened | No, assumes info you have |
| Cost per use | About 2 minutes to search | About 10 minutes done properly |
| Works without any tool | No | Yes, pen and paper |
The honest failure mode of Screenpipe, beyond the obvious privacy tradeoff of recording your own screen and everyone on your calls, is that it can make you feel more rigorous than you actually are. A searchable archive of what happened is not evidence you reasoned well about what to do next. I still run it, the factual-recall use case earns its $25 a month on its own, but I never skip the premortem before a decision that actually matters, because Screenpipe answers what happened and Kahneman's tools are the only thing I've found that reliably improves what should happen next.
Sources: Screenpipe launch thread and pricing, github.com/screenpipe/screenpipe (July 2026); "Launch HN: Screenpipe (YC S26)," news.ycombinator.com, July 2026; Daniel Kahneman, Thinking, Fast and Slow (2011); coverage of the Limitless/Meta acquisition, December 2025.
Related: How to Find Your Passion · Best Self-Improvement Books · How to Make Better Decisions · AI Coach App — Building It in 8 Hours