From securing trillions in banking to fixing AI's memory problem

👋 Hi Product Hunt. I'm Rav.

For 20 years my job was securing systems that move trillions of dollars in global banking. 6 patents, a lot of time spent on one question: why do trust models that pass design review break at scale?

The answer, over and over, was memory. The systems assumed a verified record existed, what's currently true, who committed to what, what superseded what, and almost nowhere did it. Then I noticed my own AI had the exact same disease: a great session Monday, gone by Wednesday, re-explaining my week to a machine every morning.

So I'm building MrAgentˣ, a context layer for humans and their agents. You talk to Claude or ChatGPT like you already do; it keeps the state (dated facts, what changed, what a document proves vs what the AI concluded) so the next session starts where the last one ended instead of from zero.

I'm here to learn as much as to build, I've been reading the what-are-you-building threads and they're better than most newsletters. If you work on agents, memory, or trust, I'd love to compare notes. Launching soon, but mostly just glad to be in the room.