A founder note
Why human authority.
N. V. Karuna · July 2026 · Draft — awaiting founder sign-off
Every conversation about enterprise AI eventually reaches the same question, usually asked by the most senior person in the room: who signs? Not "who monitors", not "who is in the loop" — who puts their name on the outcome and answers for it when the auditor, the regulator, or the board asks.
The industry's stock answer is "human-in-the-loop". I have come to believe that phrase is the problem. A loop is a pipeline diagram; a human in it is a component. Components get optimised out. When the loop gets busy, the human becomes a rubber stamp — technically present, practically absent, and legally exposed.
Regulated India already has a better word for it, and it is written in green ink. A gazetted officer's countersign is not a checkpoint in a process; it is the assumption of authority. The maker prepares, the checker signs, and from that moment the outcome has an owner. Maker–checker survived every technology shift of the last fifty years because it encodes something regulation actually wants: accountability that cannot be delegated to a machine.
So we built the company on a constraint instead of a capability: agents propose, humans approve. Our agents are genuinely capable — they reconcile, draft, orchestrate, and route work across systems that have never spoken to each other. But the architecture refuses to let them approve, execute ahead of a signature, touch the audit trail, or override a person. That refusal is not a limitation we apologise for. It is the product.
The bet is simple: in BFSI, healthcare, government, and manufacturing, the AI that wins will not be the one that removes people from decisions — it will be the one that gets decisions to the right people, with the evidence attached, fast enough that automation and accountability stop being a trade-off.
If you carry sign-off authority in a regulated business, this product is being built around your signature. I would like to hear what stalls on your desk.