Use cases
Start with the work your team wants to improve.
Three illustrative jobs, each with the evidence, the change, the people involved and how you'd measure it. The billing examples are fictional.
- 01
Investigate recurring customer confusion
- The problem
- Customers keep asking why their invoice changed after switching plans. Each conversation gets answered, but the questions don't stop.
- Example evidence
- A group of similar conversations with their sources, how many mention a plan change, and what share of conversations the finding covers.
- The intervention
- Clearer invoice wording and a help article, before any automation is considered.
- Human involvement
- A billing lead owns the change. New wording and articles are reviewed before they are published.
- How to measure it
- Follow-up contacts about invoice changes, compared with a baseline set before the change.
- Capabilities
- Customer intelligence
- 02
Help a representative handle a complex question
- The problem
- A refund question needs the refund policy, the account's billing history and some judgment, and the answers sit in different tools.
- Example evidence
- The relevant policy, the customer's billing records and similar resolved cases, each linked to its source.
- The intervention
- A suggested reply and a proposed action inside the helpdesk, for the representative to review.
- Human involvement
- The representative edits, approves or rejects every suggestion. Nothing is sent without them.
- How to measure it
- How often suggestions are used, how often they need editing, and whether the customer has to follow up.
- Capabilities
- Agent Assist
- 03
Coordinate a service request that needs judgment
- The problem
- Refunds above a limit stall between support and finance. Customers chase, and nobody can see who owns the next step.
- Example evidence
- The case context, the policy that applies and any earlier refunds on the account.
- The intervention
- A permitted AI agent handles in-limit cases. Larger ones go to the right approver with their context, and are tracked until settled.
- Human involvement
- A supervisor approves anything above the limit. Finance confirms settlement. Every step has an owner.
- How to measure it
- Time from request to settlement, and the number of cases waiting longer than an agreed limit.
- Capabilities
- AI agents · Workflow orchestration · Operational oversight
Bring a workflow from your team.
Tell us which customer work your team keeps chasing. We'll show you how didii fits it. Early access is free.