Refund and exception decisions, made consistent
The same case gets a different answer depending on who is on shift. Solon makes the good call the standard call.
The drift problem
A carrier is late and the customer asks for a refund. One manager approves it. Another manager, on another shift, denies the identical case. Neither decision is wrong in isolation, and the business now has two precedents and no way to tell which one its policy actually holds.
What changes when precedent becomes policy
- The first override is recorded with its context.
- Repeats cluster, and Solon drafts the rule the team is implicitly already following.
- The rule arrives as a pull request, reviewed like any other change.
- Merged policy resolves the next case the same way, whichever manager is on shift.
What the team sees in week one
Escalation trends in a weekly review, the first proposed rules, and a policy repository that starts to read like the team's actual judgment instead of a document nobody opens.
Start from the runbook you already have
Most teams keep their exceptions in a spreadsheet, a help doc, or a manager's memory. Solon does not need a clean starting point. Point it at the workflow and let the overrides teach it.
Run a 30-day pilot
We deploy Solon against your refund or exception workflow. You bring reviewers; we bring the loop and the metrics.
Start a 30-day pilotFree during the pilot · exit anytime · merging stays a human action
Keep reading
- AI decision governance: AI decision governance governs whether a decision was right, not just what an agent can do. Solon learns from human overrides and turns them into versioned policy.
- How Solon turns overrides into policy: Capture every decision, cluster repeated overrides, draft a policy pull request, and keep the merge a human decision. The loop, step by step.
- Frequently asked questions: What AI decision governance is, how Solon learns from human overrides, what the pilot includes, and what stays a human decision.
- Run a 30-day pilot: We deploy Solon against your refund or exception workflow. You bring 2 to 4 reviewers; we bring the loop, the training, and weekly metric reviews.