> ## Documentation Index
> Fetch the complete documentation index at: https://docs.triplesession.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Smart selection hints and per-scorecard languages

> Tune how AI Coach routes calls to each scorecard with smart selection hints and AI feedback, and set scorecard and insights output languages independently.

<div style={{display: 'inline-flex', alignItems: 'center', gap: '0.5rem', fontSize: '0.875rem', color: '#6b7280', lineHeight: 1, marginBottom: '1rem'}}><span style={{lineHeight: 1}}>June 5, 2026</span><Badge>New</Badge></div>

Smart Scorecards now give you finer control over which calls each scorecard is applied to, and scorecards can produce evaluations and insights in the language your team works in.

## Smart selection hints

Each scorecard's **When to use this scorecard** description is what AI Coach reads to route the right calls to it. Now you can back that description with optional structured signals — smart selection hints — that sharpen matching when the description alone is not enough.

<Frame>
  <img src="https://mintcdn.com/triplesession/Xrax255VAgiCKVSp/images/ai-coach/scorecard-edit-smart-selection-hints-overview.jpg?fit=max&auto=format&n=Xrax255VAgiCKVSp&q=85&s=65c2193575acd8bcc58d132df0395296" alt="Smart selection hints — Call stage, Participant type, Minimum call duration, Required signals, Example phrases, Disallowed scenarios, and Often confused with" width="2346" height="1834" data-path="images/ai-coach/scorecard-edit-smart-selection-hints-overview.jpg" />
</Frame>

* **Call stage** and **Participant type** — the stages these calls happen in and who joins from the customer side.
* **Minimum call duration** — skip calls shorter than a set length.
* **Required signals** — describe, in plain words, what has to happen on the call for the scorecard to fit.
* **Example phrases** — short quotes people actually say on these calls.
* **Disallowed scenarios** — situations where the scorecard should never be used; AI Coach skips it automatically when a call matches one.
* **Often confused with** — point to similar scorecards so AI Coach works harder to tell them apart.

Per-call context like provider, attendees, and duration is detected automatically, so you only fill in what the matcher can't infer.

## Get AI Feedback on your description

A new **Get AI Feedback** button reviews your description and hints, checking whether you've defined the call type, included essential qualifiers, and stated what's excluded. It flags gaps and proposes specific fixes — for example, setting a minimum duration or adding a participant type — that you apply in one click.

<Frame>
  <img src="https://mintcdn.com/triplesession/Xrax255VAgiCKVSp/images/ai-coach/scorecard-edit-ai-feedback-suggestions.jpg?fit=max&auto=format&n=Xrax255VAgiCKVSp&q=85&s=b4c5dcd4969ef435cc3fc49174764351" alt="AI feedback panel listing suggested edits for Minimum call duration, Participant type, and Required signals, each with an Apply button" width="2250" height="1728" data-path="images/ai-coach/scorecard-edit-ai-feedback-suggestions.jpg" />
</Frame>

## Scorecard and insights languages

Scorecards now carry two independent language settings:

* **Scorecard language** (Details tab) — the language used for scorecard prompts, criteria, and generated evaluation text.
* **Insights language** (Insights tab) — the language used for generated insight titles and prompts.

<Frame>
  <img src="https://mintcdn.com/triplesession/Xrax255VAgiCKVSp/images/ai-coach/scorecard-edit-insights-language.jpg?fit=max&auto=format&n=Xrax255VAgiCKVSp&q=85&s=deb2002390309ef1c962e557337b450e" alt="Insights language selector on the scorecard Insights tab" width="2350" height="1354" data-path="images/ai-coach/scorecard-edit-insights-language.jpg" />
</Frame>

Because they're separate, a team can score calls in one language while producing insights in another. Both are set per scorecard and apply to new meetings reviewed with that scorecard.

See the [Scorecards](/ai-coach/scorecards) guide for full details on configuring descriptions, hints, and languages.
