Settings
How much Darwin may do on its own, and where it reports back.
Autopilot
Checking…
Off. Darwin keeps everything it has learned and changes nothing until you switch it back on.
No test running right now. The bar is fixed, so every change is held to the same standard.
Simulated shoppers
About 12 people and 3 AI shoppers every couple of seconds, so Darwin has something to learn from. Tagged simulated, never mixed into real numbers.
Your store
- Not connectedYour GitHub accountChecking…
- Demo or dry runDarwin's brainBuilt-in rulesChecking…
Who Darwin tests for
Each fix is judged only on the shoppers it can reach, so the others can't add noise to the call. Darwin works this out from what the fix changes.
- PeopleChanges to the pages shoppers see. AI shoppers never load them.
- AI shoppersChanges to what the store API tells agents. People never see them.
- AllEveryoneChanges both can feel, like a free-shipping threshold that shows up in an agent's landed price.
Automatic, so there's nothing to switch here.
Your Grok teammate
A Grok bot reads Darwin every hour and messages you when there's a call to make, like “your new checkout is winning at 91%: want me to ship it?” Say yes and it ships, through the same API below.
See exactly what the bot would send you right now, from Darwin's live numbers.
- 1Every hour, it reads Darwin's briefing.
- 2If a test is ready or clearly losing, it messages you with one question.
- 3You answer yes, and it ships or stops the test, then tells you what happened.
The bot fetches the briefing and forwards the headline as-is. Simulated numbers are labelled.
curl -s https://your-darwin.app/api/briefing \ -H "Authorization: Bearer $DARWIN_ADMIN_TOKEN"
Keep DARWIN_ADMIN_TOKEN in the bot's secrets. Read the full guide
Start over
Puts the store back to the original page, Gen 0, and forgets every recorded visit, test and issue. Your GitHub and Whop connections stay.
Demo mode
Runs the whole loop in your browser, no server. Handy for showing Darwin off.