Your store agent

AI shoppers ask it what you sell, and buy through a checkout link that credits them.

A/B tests on your agent

Darwin changes one thing about how the agent pitches at a time and keeps it only if more conversations end in a payment. The pitch today: a plain list of offers.

Facts up front
Answer the buyer's objections: instant access, cancel any time.
One best pick
One recommendation with a reason, not a menu of three.
Structured buy instructions
Exact offer ids and prices for agents that read data.
Upsell the yearly plan
Lead with the biggest plan. Bigger orders, maybe fewer.

How Darwin calls a test

conversations per arm: 100+; conversations per arm
payments before any call: 30+; payments before any call
chance better to ship: 97%; chance better to ship
to stop it: 3%; to stop it

No test running. Turn on Autopilot, or press Test it on a lever to start one.

Agent sales

–

of agent conversations end in a payment

–

Revenue

–

Paid

By agentpaid · higher is better

No agent has shopped yet. Run a simulated buyer, or connect your own agent.

Connect any AI agent

Shoppers' agents talk to your store here, get real offers, and buy through a checkout link that credits them.

and your own

https://www.usedarwin.app/a2a/whop

Agent card (A2A v1.0 and v0.3)

Try it from a terminal

curl -s https://www.usedarwin.app/a2a/whop -H 'content-type: application/json' -H 'x-agent-name: my-agent' \
  -d '{"jsonrpc":"2.0","id":1,"method":"SendMessage","params":{"message":{"role":"ROLE_USER","messageId":"m1","parts":[{"text":"coaching under £40 a month"}]}}}'

What it sells

Demo offers until a Whop business is connected.

Talk to it like a buyer agent

Every message goes over the real /a2a/whop endpoint, exactly as an outside agent would send it.

Ask what's for sale, then say “buy the first one”. Or run a simulated buyer agent and watch it shop.