Overview
While we were deep in customer interviews for the observability work, my co-founder came across a white paper from Auth0 about identity management for AI agents, and it stopped us both cold.
The problem was enormous. Companies were deploying AI agents everywhere, but nobody had figured out how to control what those agents could access, what they were allowed to do, or how to shut them down if something went wrong. For big companies especially, that was not a nice-to-have. It was a blocker. We decided that was where Bloom needed to be.
Contribution
Product + Engineering
Team
Ruthvik Jonna
Aditya Naidu
Date
July 2025 - May 2026
Process
We had the technical foundation from the observability work, so the shift felt more focused than starting from scratch. The product was an access control platform built specifically for AI agents. Think of the way companies decide which employees can access sensitive systems. We built that, but for AI agents instead of humans.
Again, we did what we always did inside Bloom: cold outreach, customer interviews, and building in every gap in between. If we were not building, we were talking to people. Then about a week and a half in, I got a message I did not expect. I had cold emailed a VP at Salesforce about the product, and he forwarded it to a senior VP at Okta who wanted to get on a call with us. A week and a half after we started building.
Outcome
That call was just the beginning. Pilot interest started coming in. People were reaching out to us instead of the other way around. We applied to Y Combinator and found out we were in the top 10% of applicants for the Fall 2025 batch. We did not get in, but something had clearly shifted.
Then school started back up and competing with well-funded platforms as full time students was not something we could sustain. Eventually we stepped away from working on that iteration of Bloom, three and a half years after that first text message.
But something tells me this story is far from being over ;)