Any creative ways to transfer context between AI models?
did anyone find best ways to not repeat yourself to every AI?
gravez
Show HN: Kortyx – Personal memory layer for every AI agent
Hey HN I’m Dima.I built Kortyx after a failed attempt to find product-market fit with an AI product-analytics startup.When it was time to pivot i was looking for other ideas and figured that i constantly struggled with getting users to provide "ideal" context to my agents. it resulted in weak prompts, degraded agent performance, etc etc.So i thought what is instead of asking every user to...
Show HN: UserWatch – AI product analyst. Instant dashbords, AB tests, AI replays
tldr: it's essentially Cursor but for Product Analytics. AI that sets up your PostHog.Hey, I'm Dima, a former PM who led product redesign and had a constant analytics hangover, and now a solo SF founder who only has time either to ship a new product or recruit users. Analytics is always lagging behind and piling up into an annoying debt.I was also reflecting on product management as a job in...
Re:Ask HN: Cursor is productive for an hour, then burns my application...
on Friday June 06, 2025 @03:52PM (#44204364)
oh, gotcha. Interesting approach with .md files! I also have problems with agent getting carried away and starting to create documentation, test files, etc. mostly notice this with Claude 4.
It's just cursor's system prompt problem. they just need time to "tame" the model after release.
for now, I just make sure that in every chat thread i have "DO NOT WRITE ANY DOCUMENTATION OR TEST OR ANYTHING THAT WASN'T EXPLICITLY ASKED. STAY LEAN"
But i sort of reached the point where I don't mind claude going off the rails a bit. Like restricting it with .md and constantly updating those guardrails sounds like more of a burden than help.
its just the prompt problem. try reading the chat and every time you see it doing excessive shit stop the chat and slap it on wrist saying "never create .env i already have it, you just don't have access, etc".
also, sounds obvious, but don't forget to create new conversations often. the "ignore a files it's just created" sounds like context window overload. 200k window for a new model sounds like a crime from anthropic
Re:How to post when no one is reading
on Monday June 02, 2025 @11:32AM (#44159891)
Yeah, agree. The self-pressure to write a good post for others, for lead-gen, for brand awareness, all take away from "things you like".
Something that's been working for me lately is to choose the topics where you have something to say. It's a bit broader than the things you like and allows you to just react to an inner spike to respond. Helps train the muscle for writing
Re:Ask HN: Cursor is productive for an hour, then burns my application...
on Monday June 02, 2025 @11:26AM (#44159846)
can you explain a bit more what do you mean by burning down? and what do you use .md files for? Documenting the code?
Show HN: UserWatch – AI that runs your PostHog(dashboards, A/B tests, AI replays
Hi HN, Dima here - I'm building prompt‑driven product analytics. Essentially, AI that sets up your PostHog (dashboards, A/B tests, replay analysis).Why working on this So I'm a former PM, and product analytics seemed like only semi-"hard", transferable skill one can have. But the problem is that for every company, you still have to find the key metrics anew and the process of finding the right...
Show HN: UserWatch – AI re‑watches your session replays and labels every dropoff
Hi HN,I used to be the PM responsible for onboarding and activation. After hundreds of hours binge‑watching session replays and running user interviews, I realised we were still answering “Why did this user ghost after the trial?” on vibes.So I built *UserWatch*, an AI system that:1. Captures plain‑vanilla session replays (drop one script tag). 2. Re‑watches every session and tags each exit...