AI product manager · Bay Area

Miya is my personal AI agent. Rahat is the platform underneath it.

At home I build and run my own agents, with checks in the path, a harness every answer runs through, and a record of every decision. This is where I show the system, write about what it takes to trust agents, and ship a demo every two weeks.

See the system → Read the writing

Rahat end to end: a message travels from Telegram through Miya's runner and the routing ladder to the agent that answers it, with the checks, the control plane and the memory around it — A message moving through Rahat, from Telegram to the agent that answers it.Open the live page →

7agents
3front doors
4,209tests
72real asks in the eval set
7,408decisions logged

Counts as of Oct 17, 2026. Every one is in the repo.

Start here

Three ways in

The system

Rahat

Seven agents sharing one memory, one rulebook and one record of every decision. What's running, what isn't, and the code.

See the system → The series

Where do we eat?

My food agent wasn't ready for our family trip, so I made the trip its test. Six short episodes on what it takes to ship an agent.

Read the writing → Every two weeks

Demos

One short, dated demo of something that shipped, with what's real and what isn't. No mockups.

Watch the demos →

A rule in a prompt is a request. A check in the path is a guarantee.

An agent can be fluent and wrong at the same time, and you can't tell from reading it. Run more than one and they contradict each other, act at moments they shouldn't, and remember the wrong things. Those problems don't live in the model. They live in the layer underneath it. That layer is what I build.

New