... everyone. Raffay here. Building Finly AI: voice/text expense entry, receipt & PDF import, and spending insights that are actually useful instead of another spreadsheet with extra steps. Background's more backend/infra than finance, so the funny part is the AI side was the easy build. The trust and security work that comes with touching people's money data has eaten way more of my time than expected. No big manifesto, just building and trying to get it right before ...

... even come up with a solution. And then test everything on real customers, not in your own head. Comment from Raffay Sajjad(@raffay_sajjad): The email signup one got me. Had a decent list built up before finly's first version even worked properly and read it as validation. Actual activation once people could get in was way lower than the signup number implied. Signups measure curiosity, not whether someone's willing to change what app they open every morning ...

... user___1702026e6f946759d1d798a): This is also one of the first things investors look for. A polished demo can attract attention, but if someone cannot quickly understand who has the problem, how painful it is, and why the current alternatives are inadequate, the conversation hits a dead end. Based on my own experience and a friend who works at 1752vc, the strongest pitches often make the problem feel inevitable before they spend much time explaining the product. Comment from Elias(@elias ... ... Comment from Anna Ludwinowski(@anna_ludwinowski): @getosmo  "The receipt IS the scope" - YES, that's a sharper way to collapse messaging and product into the same thing. Most people treat those as two separate problems (what we say vs

... Context Engine for my Models 4. Replicate - Content Generation & Training Image & Video Models 5. Resend - API managed emails Also, recently - its Klipio, My own tool for Static Image ad creatives from competitor ads. Comment from Justin Verlin(@justin_verlin): Claude Code VS