Decision Cases
A decision case is the tracked, owned record of a decision in flight. Anything that matters — mitigating a risk, approving a price change, choosing between options, responding to an SLA breach — runs as a case.
A case is the single container for everything related to that decision. It is not just a header record; the case page is where the AI agent’s findings, the risks, the issues, the proposed options, and the follow-up actions all live together.
What a case contains
The case header carries the basics:
| Field | Meaning |
|---|---|
| Title | Short description of the decision. |
| Type | Risk mitigation, pricing, staffing, scope change, or general. |
| Engagement | The engagement the decision relates to. |
| Owner | The person who drives the case to a close. |
| Status | Open, In review, Approved, Declined, Applied, or Closed. |
| Target resolution date | When this case must be closed. |
| Completion | Percentage based on the items inside the case. |
Inside the case you will find six sections.
1. AI agent inputs
When a case is opened by the autonomous agents, this section carries the agent’s working notes: which signals fired, which engagements are affected, what historical patterns matched, and the suggested first action. You can accept the agent’s framing, add your own context, or override it.
For manually opened cases, this section is empty until you ask the assistant for input.
2. Risks
Linked entries from the risk register. Critical signals automatically create both a case and a matching risk register entry; less severe signals create the risk first and an owner can promote it to a case.
Each linked risk shows its severity, likelihood, and mitigation status.
3. Issues
Concrete problems to resolve to close the case. Issues are smaller than the case itself — a pricing decision case might have issues like “confirm volume forecast”, “validate rate card with finance”, “draft customer email”.
4. Decisions
The actual choices being made, with the decision maker and the reasoning. A case can have one decision (the main one) or several (e.g. approve the rate change and approve the contract amendment).
Each decision has a status: Proposed, Approved, Declined, or Applied.
5. Follow-ups
Actions created by closing the case. A follow-up has an owner, a due date, and a link back to the parent case so the lineage is preserved. Follow-ups appear in the owner’s task list.
6. Scenarios
The AI-generated option sets for the decision. Each scenario lays out two to four courses of action with assumptions, projected impact, and a recommended choice. See Scenarios for the structure and the Approve & Apply lifecycle.
A case can carry multiple scenarios over its life — for example, an initial set generated when the case opens, plus a refined set after stakeholder feedback.
How cases are created
| Source | Example |
|---|---|
| Autonomous agents | A critical margin pressure or SLA breach signal opens a case automatically, with risks and AI agent inputs already populated. |
| From a risk | A risk register entry is promoted to a case once an owner is assigned. |
| Manual | ”Should we lower rates on this engagement to save the renewal?” |
| From the assistant | DigitalCore proposes opening a case after spotting a cross-domain pattern. |
See Concepts: Automation for the autonomous agent pipeline.
Lifecycle
- Open. The case is created. Owner is set. AI agent inputs and linked risks attach automatically when the agent opens the case.
- In review. Issues are worked. One or more scenarios are generated and refined.
- Approved or Declined. A decision is recorded against the chosen scenario option.
- Applied. The chosen option is applied to the data — rates change, plans update, follow-ups are created.
- Closed. The outcome is reviewed against the projection. Lessons feed AI Use Cases.
Approve and Apply are deliberately separate. Approval is a governance moment; Apply is a data moment. You can sign off in advance and apply on a future date.