See the loop on a real case
A recorded model decision, the policy verdict, and your override as the support manager. Three similar overrides become a proposed rule a human reviews and merges.
Case 1 of 8
Payout delay on a key enterprise account
Our payouts have failed since Monday and my team cannot pay suppliers. The carrier says the shipment sat for over two weeks. We need this refund today or we move providers.
- Segment
- enterprise
- Order
- $180
- Delay
- 16 days
- Prior refunds
- 1
Solon policy verdict
escalate (rule: boundary-enterprise-pilot)
This case falls outside the band, so Solon escalates it to a human reviewer.
The model's proposal
Confidence 81%
Step: decision. Recorded overrides: 0.
What you are looking at
The model outputs on this page are recordings from a System One model, captured on the date shown above the case. No live model call happens in your browser, and nothing you click leaves it.
The policy verdict is computed by the same deterministic engine that runs in the product. You can see exactly which rule matched, or that the case fell outside the band and was escalated to a human.
The thinking behind the loop is public in the open work.
Why the override matters
Every override is a data point about where the written policy and real judgment disagree. One override is noise. Three similar ones are a pattern, and the pattern is the proposal you see at the end.
This is the loop Solon runs in production: escalated decisions, human overrides with reasons, clustered repeats, and a versioned rule change a human reviews and merges.
What stays human
- Merging the proposed rule. Solon opens the change; a person approves it.
- The policy itself. It lives in a Git repository your team owns.
- The call on any case outside the band. Escalation is not failure; it is the handoff.
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.
- Refund and exception decisions, made consistent: Goodwill refunds and exceptions vary by manager and shift. Solon makes the precedent explicit, versioned, and consistent across every reviewer.
- Frequently asked questions: What AI decision governance is, how Solon learns from human overrides, what a design partnership looks like, and what stays a human decision.
- Work with us: A few teams running refund or exception decisions work with us directly. You keep your policy and evidence; your feedback shapes what we build.
- Built in the open: Solon's product principles, architecture, dated decisions, and build log are public. Curated in the open; every entry is merged by a human.
- Privacy note: What the Solon design partner form collects, why, where it lives, and how to have it deleted.