The Overview is the summary you get after each call. Out of the box it covers what was said. Customised, it answers the questions you actually ask after every call of that type, which is a very different thing.
What customising it means
You do not rewrite the Overview by hand. You tell Claap which AI Fields should feed it, and those fields are then reflected in the summary.
An AI Field is a question asked of every call, for example the customer's needs, who holds authority, or what the decision process looks like. Attach those fields to your Overview and each summary comes back structured around them rather than around whatever the call happened to cover.
The practical effect: instead of reading a neutral recap and hunting for the decision process, you get an Overview that leads with it, every time.
Two ways to set it up
For yourself, in Personal Settings, then Meeting preferences, on the Automation preferences tab. Internal and external meetings are configured separately, each with its own Generate the following AI fields. That separation matters: the questions worth asking on a customer call are rarely the ones worth asking on a team sync.
For everyone, in Workspace Settings, then Admin automations. Create an automation, set a trigger such as Meeting type is External, and add the action Auto-personalize summary. You then select the AI fields that should personalise the Overview summary, for example Customer Needs, Authority and Decision process.
Admin automations also have an Enforce automation option, which prevents members from overriding the rule with their personal preferences. Use it when consistency matters more than individual choice, for example when the Overview feeds a CRM field or a report.
Choosing the right fields
Start from what you write down after every call anyway. If you always note the next step, budget and who else is involved, those are your three fields.
Three or four is usually right. An Overview personalised with a dozen fields stops being a summary and becomes a form, which defeats the point.
Prefer questions with a stable answer shape. "Was a next step agreed" produces something comparable across two hundred calls. "How did the call go" does not.
And remember each field costs one credit per meeting, so the cost is fields multiplied by call volume, not the number of fields.
Check it on a real call
Run it on an actual meeting rather than a clean test. If a field comes back empty or vague, the question is usually too broad rather than the call being unusual.
Adjust the field, not the Overview. The Overview reflects the fields, so that is where every change belongs.