Workspace intelligence score

Every question you ask, every correction you make, and every reveal you certify teaches Orion something about how your business works. The Workspace Intelligence Score makes that accumulated knowledge visible. As the score rises, Orion needs less hand-holding to get your metrics right — and you can see precisely what it has learned.

What the score measures

The score is a number from 0 to 100, aggregated from five sources of workspace knowledge. Each source contributes points up to a cap:

The caps add up to a maximum of 100. A high score means Orion has a rich, governed understanding of your business — not just that you've used it a lot.

Levels

Your score maps to one of four levels:

The intelligence widget

The home page shows a compact Orion Intelligence widget:

• The current level badge (e.g. Trusted)
• The score out of 100, with a progress bar
• A plain-language summary, such as "Orion has learned 47 things about your business"
• A See what Orion knows → link that opens the full breakdown

What Orion knows

Click See what Orion knows to open the expanded panel. It shows:

• A breakdown by component, each with its count and points, drawn as mini bars — so you can see which source is contributing most.
• Highlights — a human-readable list of the most important things Orion has learned, phrased as "Orion knows that…". For example:

> Orion knows that…

> - Revenue is your primary KPI

> - CAC calculation excludes brand spend

> - Order counts always exclude cancelled orders

• A Teach Orion more action that opens Orion pre-filled with "Remember that…", so you can add a fact directly.

How to improve your score

Because each component is capped, the fastest way to raise a low score is to broaden across components rather than maxing out one:

• Certify your reveals (5 pts each, up to 25) — the highest-value action. Certify metric definitions in Forge.
• Correct Orion when it's wrong (2 pts each, up to 20). Each correction is stored in memory and applied to future answers.
• Teach Orion key metrics (3 pts each, up to 30). Tell Orion your canonical definitions and gotchas via "Remember that…".
• Set up KPI monitoring (2 pts each, up to 10) so Orion tracks snapshots of your most important numbers.
• Keep asking questions (1 pt per 10 queries, up to 15) — this accrues naturally with use.

Corrections and remembered facts feed Orion's workspace memory, which is loaded silently at the start of every session.

For admins (Lumen)

Admins get an expanded intelligence panel in Lumen, on the queries page below the Corrections panel. It includes the full component breakdown and the ability to edit workspace memory directly — add, fix, or remove the facts Orion relies on.

API

The score is exposed through a single workspace-scoped, authenticated endpoint:

A typical response:

{
  "score": 73,
  "level": "expert",
  "components": {
    "key_metrics":      { "count": 8,  "points": 24 },
    "corrections":      { "count": 6,  "points": 12 },
    "certified_reveals":{ "count": 4,  "points": 20 },
    "queries_run":      { "count": 90, "points": 9 },
    "kpis_monitored":   { "count": 4,  "points": 8 }
  },
  "highlights": [
    "Revenue is your primary KPI",
    "CAC calculation excludes brand spend"
  ],
  "last_updated": "2026-06-09T08:00:00Z"
}

Limitations

• The score is cached for about 10 minutes per workspace. It is not real-time — actions you take (a correction, a new certified reveal) may take a few minutes to be reflected.
• It measures governed knowledge, not raw activity. Running thousands of queries alone can't push you past the queries cap of 15 points. To reach Trusted, you need certified reveals, corrections, and remembered metrics.
• The score is workspace-scoped. Each workspace builds its own intelligence; knowledge does not transfer between workspaces.
• A high score does not guarantee correctness. It reflects how much Orion has learned, not that every learned fact is still accurate. Review workspace memory periodically in Lumen and remove anything stale.