The most-covered Indian consumer AI pre-seed of May isn't actually a consumer AI company.
I've been reading Banza since their $1M pre-seed two weeks ago — Campus Fund led, ZK-proof stack, founded in January. Three things the press missed.
First, the team. Suraj R. Mulla, the CBO, was at Privado ID (formerly Polygon ID) and LedgerFi IT Solutions. The “Avalanche” in the cap table is almost certainly the crypto Avalanche, not a generic fund. The stack is zero-knowledge proofs and blind compute — verifiable credentials by another name. Drops are a token by another name. When Web3 fundraising dried up, identity teams needed a new wrapper. “Personal AI Twin” let them keep the architecture, drop the crypto language, and raise from Campus Fund instead of a16z crypto.
Second, the wedge. The Play Store listing, not the funding press, says Banza starts with food because it's the decision people make most often. Food is the right wedge — high frequency, low stakes, the cleanest training surface for cross-app behavioural inference. If Banza can't make food work, the other categories never matter.
Third, the cost-flow inversion. Every other consumer AI in India asks the user to pay — ₹999/month, freemium, ads. Banza pays the user to surrender connection consent. Category-defining, not category-copying. Datacoup tried this in the US and died. Brave made it work for browsers but never apps. Worldcoin is doing it with biometrics at $100M+ scale. India hasn't seen a serious attempt at this until Banza.
What I'd watch in 12 months is whether the next round discloses engagement metrics. If they raise $5M+ and quote DAU/MAU, the Twin is earning its keep without Drops.
If they raise flat with no numbers, the Web3 rebrand is what got them to pre-seed and won't get them to seed.
If you're building in Indian consumer AI and read Banza differently, I'd take a counter.