The masthead above your Clinic Health Index
At the top of your Dashboard, above the Clinic Health Index pillar breakdown, sits a masthead labeled AI summary with a small sparkle icon next to it. It holds one or two plain-language sentences describing how your clinic is doing right now — something like "Your clinic is in good shape this week, though a few overdue recalls are worth a look." It's meant to be the first thing you read on the Dashboard, before you scroll into the pillar-by-pillar detail underneath it.
Because it sits in that prime position, it's worth being precise about exactly what it is and isn't. This article covers the mechanics in more detail than you'd normally need — but if you manage a clinic, understanding exactly what feeds this sentence is the difference between reading it as a helpful nudge and mistaking it for an authoritative number.
- Exactly what data is sent to the AI to generate the summary — and what's deliberately withheld
- Why the AI can describe your clinic's state in words but structurally can't state a number
- What the fallback sentence looks like when the AI summary isn't available
- Why the "AI summary" label stays visible even when a fallback sentence is shown
What's actually sent to the AI
It helps to think of this as a two-stage pipeline rather than one step. Stage one has nothing to do with AI at all: ClinyPal computes your Clinic Health Index score, its band (for example, "Good" or "Needs attention"), and your clinic's top action items using its normal, deterministic reporting logic — the same kind of calculation that produces every other report in the product. That stage finishes completely before the AI is ever involved, and its output would be exactly the same with or without AI in the picture.
Stage two hands a narrow slice of that output to the AI: the already-computed score, its band, and the titles of your top action items. Nothing else. The AI never receives your appointment counts, revenue figures, patient totals, or any of the underlying numbers that fed the score calculation — only the finished score and a short list of item titles like "3 overdue recalls" or "2 forms pending signature."
| What's sent to the AI | What's never sent |
|---|---|
| The clinic's already-computed health score and band | Raw appointment counts or revenue figures |
| The titles of top action items (e.g. "3 overdue recalls") | Any other underlying metric behind the score |
| An instruction to describe clinic state in one or two sentences | Permission to state or invent a specific figure |
When the AI summary isn't available
If the AI call fails, times out, or isn't configured for your clinic, the masthead doesn't disappear and it doesn't show an error banner. Instead, a canned, pre-written fallback sentence appears in its place — something generic and safely true regardless of your clinic's actual state, like a reminder that the Clinic Health Index below reflects your current standing.
This is the one detail in this category most worth internalizing: the label on the masthead never changes based on whether the sentence underneath it was actually written by AI in that moment. "AI summary" identifies the feature, not a live guarantee about the specific sentence currently in front of you.
In day-to-day use this rarely matters — the fallback is written to be true and reasonable in any state, so it's never misleading, just less specific than a genuine AI sentence would be. But if you're ever unsure whether you're looking at a fresh AI read of your clinic or the generic fallback, the practical tell is repetition: a fallback sentence tends to read the same way visit after visit, where an AI-generated one shifts as your action items change.
How this fits with the rest of your Dashboard
The AI summary is a masthead, not a replacement for the Clinic Health Index itself. The pillar-by-pillar breakdown underneath it — the actual scores, bands, and action items — is computed the same way and shown the same way regardless of whether the AI sentence above it loaded successfully or fell back to the canned message. Nothing about the AI layer changes what the Health Index itself says.