In short
- Tests written for you.OneType turns your pages into questions a real visitor would type, plus questions your content does not cover.
- One clear score.A number from 0 to 100 with a plain verdict: Ready, Good or Needs work.
- A fix for every miss.Each failing question comes with a button to add the missing answer or open the source behind it.
- Real answers become tests.Save a good answer from your inbox as a test, so it is checked every time your content changes.
The biggest worry before putting an AI agent on your site is simple: will it get things wrong in front of customers? You should not have to find out from a complaint. The readiness score gives you an honest answer before you go live, measured on your own content.
What the score measures
A good support agent does two things. It finds the right part of your content for a question, and it says "I don't know" when your content has no answer. The readiness score checks exactly those two things and combines them into one number.
- Finds the right passage: for questions your content answers, does the right text come up among the top results the agent reads?
- Says "I don't know" when it should: for questions your content does not cover, does the agent hold back instead of guessing?
- Finding the right passage counts for three quarters of the score, holding back for the rest.
- Ready means 85 or more, Good means 70 to 84, anything lower needs work.
If none of your tests are out-of-scope questions, the score never reads Ready, because the agent's ability to hold back was never checked.
Test questions written from your content
You do not have to write a test plan. After your content is read, OneType samples passages from it and writes one question per passage in the words a visitor would use, without copying phrases from the page. A second check throws out questions that are vague or not fully answered by the passage.
It also writes questions that sound like they belong to your business but that your content does not answer. Those are the ones where a careful agent should say it does not know. When a large part of your content changes, the generated set is written again.
Fix what fails in one click
The Readiness card sits on each agent's Knowledge tab. Under the score you see what to fix: up to five failing questions, each marked "Answer not found" or "Should say I don't know". Next to each one is the fix. Add Q&A opens a saved answer form with the question already filled in. Check source takes you to the page the answer should have come from.
Click Run again to measure after your changes. The card updates on its own while the check runs. The dashboard also shows each agent's score and verdict, so you can see at a glance which agents are ready.
Turn real conversations into tests
Under any agent answer in your inbox, click Save as test. The visitor's question becomes a permanent test that expects the same sources the answer cited. Questions you replay from past tickets can be saved the same way. From then on, those questions are checked with every update, which is how your best answers stay good.