How old is your AI agent?

Not a trick question. Not testing if you've graduated to using agent harness architecture. But if a personal agent is going to shop, book, and transact for you, someone downstream needs to know things about you — age, location — or the whole idea breaks the first time it tries to buy something age-restricted or cross a state line.

🍷 Age. Most commerce is age-gated by law. Today that's a checkbox a human clicks. An agent can't click a checkbox honestly — it needs to prove the human behind it is old enough without exposing a birthdate or ID. In the Doot white paper (India's proposed national citizen-AI-agent architecture), a 70-year-old pilgrim's agent proves she's a senior citizen eligible for a rail concession — one verified claim, nothing more. Same mechanism solves age-gating for everyone else.

📍 Location. We've run this experiment before. For years, Amazon only collected sales tax in states where it had a warehouse or office — mostly just Washington. States lost billions. So the rules got rebuilt around location: click-through nexus, then South Dakota v. Wayfair in 2018, which shifted tax obligations to follow the delivery address, not the seller's home state. Location isn't metadata — it's a legal variable. And it's a business one too: US location-targeted ad spend hit $57B+ in 2025, with ~90% of digital marketers now using location data.

Personal agents won't get adopted because they're smart. They'll get adopted when they can prove just enough about the human behind them — and then forget everything else.

So, how old is your agent?