The price is the demand signal
When ventilators go scarce across the network, their price climbs before a single shelf reads empty. That climb is the signal: an early, honest read on demand outrunning supply.
Baymax sees a shortfall coming across your hospital network, negotiates a transfer between facilities, and settles it on-chain in testnet FET. It runs on its own, before a shelf ever goes empty.
9 facilities6 critical suppliessettles in testnet FET
Flip to crisis and watch the whole room warm to urgency.
Baymax exists because the hardest part of a shortage happens off the shelf: the hours of phone calls, stale counts, and paperwork between a hospital running short and the supply actually arriving.
By the time a shelf reads empty, the patient is already waiting. Counts are checked by hand on a schedule, so a shortfall is discovered hours after it became unavoidable.
Finding which nearby facility has surplus means a chain of calls, holds, and callbacks. Every handoff loses detail, and the clock keeps running while it happens.
Once stock finally moves, settling who owes whom is its own backlog. Paper trails and invoices drift, so the books trail reality long after the supply does.
One protocol runs the supply line from signal to settlement. Each step hands verified state to the next, and a person signs off before anything moves.
Claude Vision counts the physical units on every shelf camera and keeps the inventory live. External signals stream in beside it: WHO Disease Outbreak News, CDC syndromic surveillance, Open-Meteo weather and wildfire data, and regional news through Browserbase.
Baymax fuses those signals with on-hand stock and past surge patterns to project burn rate forward, so a shortage surfaces days before the shelf would run empty.
The short hospital's agent broadcasts a transfer request across the Fetch.ai network. Nearby hospital agents answer with constrained offers: quantity, price, ETA, and expiry.
Claude ranks the offers and composes a split transfer across several hospitals when no single one can cover the need. A supply exchange with an order book and market-maker pricing discovers the price live, so the settled price is itself the demand signal.
A person approves the move over iMessage, and the transfer settles on-chain in testnet FET on the Fetch Dorado testnet. If no hospital can fill it, Baymax orders from a third-party supplier. Arize Phoenix traces every step, so the next prediction is sharper.
Each node is a facility, sized by how full its shelves are. The lines between them are supply transfers, lighting up the moment Baymax moves stock from a facility with room to spare toward one that is running low.
Baymax reads real usage and live Claude Vision counts off the shelf, then projects each item forward. When the line heads for the safety threshold, it surfaces the order days early, while there's still time to act, not scramble.
Ingests. Dispensing history and per-shelf camera counts, refreshed continuously.
Projects. On-hand units forward across the week, item by item.
Flags. The exact day each item is set to cross its safety threshold.
Order 200 units of IV fluids before Friday.
One human approval in chat releases the order.
Hospitals compete for the same patients and the same scarce supplies, and when one runs short there is no line open to the others. The Supply Exchange is that line, and the price two agents settle on is the fastest read on where a shortage is forming.
When ventilators go scarce across the network, their price climbs before a single shelf reads empty. That climb is the signal: an early, honest read on demand outrunning supply.
An order book and a market maker price six critical supplies in real time. The short hospital's agent broadcasts, nearby agents answer with constrained offers, and the market finds the number. No committee, no phone tree.
Claude weighs every offer by quantity, price, ETA, distance, and expiry, then composes the plan. When no single hospital can cover the need, it splits the order across several so the shortfall still clears.
The winning plan settles on-chain in testnet FET on the Fetch Dorado testnet, and Arize Phoenix records the decision path. The money is not the point. Moving supply to where it keeps people alive is.
Baymax acts on its own to keep shelves stocked, but it never acts in the dark. Every decision is gated, recorded, and reversible up to the moment you approve it.
“I wanted the system to do less without telling me, not more. It stops and asks, and I can pull up the transaction afterward. That is what let me trust it on the floor.”
Nothing transfers and nothing settles until a person says yes over iMessage. Baymax brings you the plan, including the source, the split, and the cost, then waits for approve, order, or reject.
Each transfer settles on-chain in testnet FET on the Fetch Dorado testnet. When real settlement is enabled, the transaction reference is yours to verify, and the record outlives the conversation.
If a check fails, whether the inventory is unreadable, an offer doesn't cover the need, or a payment won't verify, it stops and tells you rather than guessing.
Facilities negotiate with each other directly. Surplus sites answer the need, offers get ranked, and the deal closes between them, with no broker in the middle taking a cut.
Baymax watches the whole network, moves supply between facilities before a shortage hits, and settles every transfer on-chain. You approve the call, and it handles the rest.
Autonomous agents · Fetch testnet · Human-approved