Put the measurement inside your own product.

The API returns the same readings the app runs: recall, category fit, buyer match, recommendations and the sources behind them. Every response carries the run date and the assistants it came from, so a reading stays checkable after you store it.

REQUEST

POST /v1/placement Authorization: Bearer sk_live_…{"site": "https://yourcompany.com", "question": "best CRM for a small design studio", "models": ["chatgpt", "gemini", "claude"], "runs": 3}

RESPONSE

{"run_id": "run_8f21c4", "run_date": "2026-09-01", "named": false, "shortlist": {"chatgpt": ["Kettleburn", "Marlowe Grey"], "gemini":  ["Sableworks", "Kettleburn"], "claude":  ["Marlowe Grey", "Verabloom"]}, "sources": ["kettleburn.com", "studiodesk.io"]}

ENDPOINTS

POST

/v1/placement

Run a question across the assistants and return who was named, in order, with sources.

POST

/v1/recall

Ask the assistants about a brand by name and return the facts they state.

POST

/v1/category

Return the category each assistant files a page or product under.

POST

/v1/match

Return the band a page lands in against a question: answers it, related, far off.

GET

/v1/runs/{id}

Fetch a stored run, including the run date and every question in it.

Runs are stored

Every call returns a run id. Fetch it later and you get the same reading, with the date it was taken.

Rate limits are per run, not per call

A measurement puts one question to three assistants several times. The limit counts those, so the number means something.

Readings expire, results do not

A stored run stays available. Whether it still reflects the assistants today is a separate question, which is what re-measuring answers.

Get a key and run one question.

Free reads work through the API too. No card until you go past them.

Get API access→View API documentation