Show every client where AI mentions them. Then move it.

One client with 10 prompts measured every week, three market scans and three measured pitches a month, and three seats. No card. Its reports carry your workspace's name and the line "Measured by Ansengine"; a paid agency plan adds clients, takes the line off and adds the quarterly review. Agency Launch is $199 a month for unlimited clients (fair use: 25 new clients and prospect workspaces a month), unlimited prompts, 10,000 AI answers measured a month and 200 AI actions a month.

That prompt is the product. AI gives a different answer each time it is asked, so no single answer is the number: every answer is graded, and each number on the report says how many answers it came from, with its likely range beside it. Below eight answers there is no percentage at all, only the counts. The raw answers export on every plan, so a client can check the grading themselves.

Every plan measures ChatGPT, Google AI Overview and Google AI Mode, each reported on its own and never blended. Three more can be added per prompt, on request, and two developer APIs on Agency Scale and custom plans. The answers are read from the apps buyers open, not only from a developer API.

On a paid agency plan, yes: your logo, your accent and your report name, and our mark stays off unless you turn it on. A report can be served from your own subdomain once you verify it, locked with a passphrase, and you can see when the client opened it.

No. Nothing publishes without an approval from your team, and an API key or connected agent you authorise can give that approval. Anything that cannot be undone takes two steps on purpose: one to arm, one to confirm. A reply in a thread goes from the brand's own connected account, disclosed, and only after approval. Outreach email goes from your own mailbox until sending from Ansengine is switched on.

Then you have the pitch. A zero prints as a zero with the count behind it, next to the competitors mentioned instead and the prompts the client is missing from. Losses print at the same size as wins on the report, which is why the wins get believed.

Ask it twice and watch the names change. One reading is an anecdote, so the free report reads 150 answers across your prompts and models, and every rate shows the count it rests on. A manual check also cannot show which pages the answer leaned on, which is where the fix is.

No. We do not post from accounts the client does not own, buy votes, sell links or edit Wikipedia for a brand. A thread gets a disclosed reply from the brand's own account, a review site gets a profile checklist and a request to real customers, and an editor gets a pitch written from verified facts.