I'm a sales guy who can't code. So I talked a software company into existence.
Hi Product Hunt 👋 I'm Ashish, founder of Atlaso.
Quick confession before anything else: I can't code. Five years in sales. I've never written a line of anything.
But for the past year I've been building — actually shipping — by talking. Literally talking. Wispr Flow turns my voice into text, Claude Code and Codex do the engineering, and I'm the guy in the middle arguing about architecture in plain English at 2am.
And somewhere in there I ran into the most ironic problem in AI:
Every morning, every one of my tools had forgotten me. My project, my decisions, my preferences — gone. I was re-introducing myself to my own team daily. Imagine hiring a brilliant engineer with total amnesia. Now imagine hiring five of them.
So I did the only logical thing: I used the AIs that kept forgetting me to build the thing that makes them remember.
That's Atlaso — one memory for every AI you use. Your decisions, preferences, and context follow you across Claude Code, Cursor, Codex, ChatGPT — every tool, every device. It's live today at atlaso.ai, there's real memory research behind it, and we're launching right here on Aug 4.
The part I'm proudest of: Atlaso remembers how it was built. The memory layer we ship is the same one that carried its own development, every single day.
Since this is my first launch ever — what's the one thing you wish someone had told you before your first Product Hunt launch? I'll take every bit of wisdom, and I'll be around in these forums long after Aug 4.
Replies
The part I left out, because the post was getting long —
Atlaso was never the plan. The plan was research.
This journey is 1.8 years old, and for the first 1.2 of those there was no product at all. Just me and my four friends: Claude Code, ChatGPT, Gemini, and Google Colab. Training runs. Benchmarks. Experiments that failed by Tuesday. Experimental products that will never see the light of day — there's a whole graveyard of them on my hard drive.
I wasn't trying to build a startup. I was obsessed with one question: can an AI's memory actually grow and get better over time, instead of just getting bigger?
Most experiments died. One refused to. The memory layer kept working — across models, across tools, across months. At some point I realized I wasn't running an experiment anymore, I was using it every day.
That's the thing that's live right now. Not an idea we rushed out in a weekend — the one survivor of 1.2 years of research and a lot of dead prototypes.
So when I say "AI brain," I don't mean it as a marketing word. I mean we grew one in a lab first. 🧪