Forums
25 years of AI, two exits, one unsolved problem
Hey Product Hunt, I am Ruban, and I have been building things with AI since before it was cool to say that.
I started my career as a data scientist at Yahoo in the early 2000s, back when "machine learning" was something you had to explain to your manager before you could use it. From there I co-founded two companies, Bixee, a vertical search engine that got acquired by Naspers, and DataRPM, a cognitive AI platform for industrial IoT that got acquired by Progress Software. In between and around all of that, I have spent 25 years obsessing over one question: why is it still so hard to turn a good idea into working software?
I watched the vibe-coding wave arrive and felt genuine excitement and then genuine unease. Not because the tools were bad, but because something important was missing. Imagine running software development straight off Slack or Teams. No Jira tickets. No Linear issues. No GitHub PRs. Just chat. Sounds scary, right? That is exactly what vibe coding does. AI makes code changes straight off a conversation thread. No traceability. No structure. No way to answer "what changed and why." - Complete nightmare to manage after that initial build.
I kept thinking: I know what happens when production systems are built without discipline, because I have been there for the cleanup. And I wanted to build the thing I wish had existed every time. This is what led me to build Avery.dev.
