Intraday Management: Copilot
Objective. Use the Copilot to speed up reading the day and framing options, while treating its output as an assist to a human decision, not an authority to act on.
The Copilot assists the read: it helps a manager understand where the day has drifted and what the options are, in plain language, so the decision is faster - but the human still decides and the limits of simulation still apply.
See it in the demo
What this screen answers
'Can something help me make sense of today faster and point me at the right move?' The Copilot is a reading and framing aid. For a duty manager under time pressure, or a newer manager still learning the screens, it can summarise the drift and surface candidate actions more quickly than scanning every interval by hand.
The task
Use the Copilot to summarise the current position and suggest where to look or what to consider, then verify its read against the underlying screens - forecast vs actual, staffing, backlog - before acting on it. Treat anything it suggests as a candidate to confirm in the what-if, exactly as you would a manager's own hunch. The task is faster orientation, not delegated decision-making.
How to read it
The Copilot proposes; the human disposes. Its summary and suggestions are decision support generated from the same data and assumptions as the rest of IDM - and any projection it quotes carries the same limits: a projection is not an executed action and not an achieved result. The classic misread is granting an AI assistant extra authority because it speaks confidently. Verify its claims against the evidence screens before you rely on them, and never present its output to a customer as fact.
Honest limits
Copilot behaviour, scope and whether it is entitled for a given customer are production- and packaging-dependent - confirm with product. It does not execute actions and its answers are only as good as the underlying feeds. Exact Copilot capabilities, phrasing and the demo figures are SME-validation-pending; the demo shows the assist model on seeded data, not a guarantee of production behaviour.
Interpretation check
The Copilot is a decision-support assist, not an authority - verify its reads and any projections against the evidence screens, and never present its output to a customer as confirmed fact.
In the field
OVO (contact centre): a newer duty manager can use the Copilot to orient quickly during a peak, then confirm its read on forecast vs actual before committing - speed without ceding the decision.
Practical task
On the demo, ask the Copilot to summarise today's position, then verify one of its claims against the matching evidence screen and write a one-line note on whether the claim held.
Self-check
The Copilot recommends opening overtime. Should the manager act on that recommendation directly?
No. It is a suggestion to verify. Confirm the drift on the evidence screens and test the option in the what-if before deciding - and execution still happens in the customer's systems.
Why does a confident Copilot answer not earn extra trust?
Its answers come from the same data and assumptions as the rest of IDM and carry the same limits. Fluency is not accuracy - verify against the evidence screens before relying on it.
A prospect asks whether the Copilot is included. How do you answer?
Confirm with product. Copilot scope and entitlement are packaging-dependent. The demo shows the assist model on seeded data, which is not a statement of a given customer's plan.
What is the honest framing of the Copilot's value?
Faster orientation and option-framing for the manager - especially a newer one. It speeds the read; it does not make or execute the decision, and its projections are still projections.
Verified vs demo. Screen is live in the demo at /#/copilot; exact Copilot capabilities, scope, entitlement, phrasing and figures are SME-validation-pending.