# AI decision record and contest > How Lumtry records every AI-assisted refund decision and lets a shopper contest it, with a person reviewing and deciding the outcome. - Canonical URL: https://lumtry.com/guides/ai-decision-record-and-contest - Last updated: 2026-09-26 Every AI-assisted refund decision writes a compliance record before the case moves on: what the model produced, whether a validator accepted it, and which injection-screen rules fired. A shopper who disagrees can contest the decision from the returns portal, and a person resolves it in the dashboard. None of this is gated by plan. ## What does Lumtry record for an AI-assisted decision? A frozen compliance record: a digest that proves what the model produced, a validator verdict of accepted, rejected or not run, every injection-screen rule that fired, and a bounded excerpt for review. The record cannot be edited once written; a later feature can only add fields with a default, never change what an earlier run already recorded. ## How does a shopper find out AI helped decide their case? The returns portal's status page shows a plain-language notice once the case finalizes, one notice per run. From that same page, the shopper can contest the decision if they disagree. ## What happens after a shopper contests a decision? The contest starts open, awaiting review. A person resolves it from the Agent Activity page's Contests to review queue, reading the shopper's reason and marking it upheld or outcome changed, with an optional note. A case can carry up to five contests before a further attempt is refused. ## Can a merchant or a reviewer rewrite what the record says? No. Resolving a contest sets it once: a repeated resolve request without a fresh idempotency key returns the same result rather than writing again. The shopper's reason and the reviewer's note stay in the dashboard; the shopper only ever sees the outcome. ## Follow a decision from record to resolution in five steps 1. **The record writes before the case moves on**: Every AI-assisted decision on a case produces its compliance record first: the digest, the validator verdict and the fired rules, so nothing about what happened is decided after the fact. 2. **The shopper reads a notice on the status page**: Once the case finalizes, the returns portal status page shows the shopper a notice that AI helped decide it. 3. **The shopper can contest it from the same page**: If the shopper disagrees, they file a contest with their reason directly from the status page. 4. **A person reviews it in Agent Activity**: The contest appears in the Contests to review queue, oldest first, until a reviewer opens it and marks it upheld or outcome changed. 5. **The shopper sees the outcome, not the reasoning**: The status page updates to show the case was reviewed and its outcome, while the shopper's reason and the reviewer's note stay on the dashboard. ## Frequently asked questions ### Can a shopper contest a decision more than once? Up to five times on the same case. A sixth attempt is refused outright, so the review queue cannot be flooded. ### Does contesting a decision undo what already happened? No. Resolving a contest records a reviewer's verdict, upheld or outcome changed, with an optional note. It does not itself reverse a refund; any further action a reviewer decides on runs through the normal case workflow. ### What survives if a case is later erased under privacy law? The bounded excerpt of what the model produced is removed. The digest that proves what was produced, and the notice and contest trail, stay in the audit record. ### Is any of this gated by plan? No. The decisions record and the contest loop are on every plan. A separate copilot that suggests policy rule fixes needs the AI rule suggestions entitlement, but recording and contesting a decision never do. ## Related - [Home](https://lumtry.com/index.md): Decide refunds by a versioned policy, send the exceptions to your team in Slack or the dashboard, and keep every step on the record. Start free. - [Guides](https://lumtry.com/guides.md): Answer-first guides to refunds, returns, chargebacks, approvals and audit trails for Shopify, WooCommerce and BigCommerce merchants. - [Slack approvals for Stripe](https://lumtry.com/guides/slack-refund-approvals-for-stripe.md): How to route Stripe refunds that need a person to Slack, approve them from the message, and keep the dashboard as the source of truth. - [Voice approvals](https://lumtry.com/guides/voice-refund-approvals.md): How outbound phone calls let an approver confirm a held refund when they are away from Slack and the dashboard, with the answer recorded. - [Return abuse screening](https://lumtry.com/guides/return-abuse-screening.md): How to score refund requests on claim history, missing returns and unusual values, and hold risky cases for a person without slowing honest customers. - [Chargeback evidence](https://lumtry.com/guides/chargeback-evidence-automation.md): How to assemble dispute evidence from orders, policy and the refund audit trail so every chargeback response starts from a complete file. - [Refund audit trail](https://lumtry.com/guides/refund-audit-trail.md): What a refund audit trail should record, why it must be append-only, and five steps to build one that holds up in disputes and audits. - [Refunds for AI agents](https://lumtry.com/guides/mcp-refunds-for-ai-agents.md): How AI assistants can look up refund cases and propose refunds over the Model Context Protocol while policy and people keep the final say. - [Retention offers](https://lumtry.com/guides/retention-offers-before-refund.md): When to offer store credit, an exchange or a partial refund before paying money back, and how a policy decides which offer fits a case. - [Marketplace refunds](https://lumtry.com/guides/multichannel-marketplace-refunds.md): How to bring store and marketplace refunds from Amazon and eBay into one queue, with a policy per channel and every decision in one audit trail. - [Simulate a policy](https://lumtry.com/guides/simulate-a-policy-before-publishing.md): How to replay a draft policy against recent decisions with no model call, and how an explainer's suggestion can land in a draft without ever publishing it. - [Canadian refund rules](https://lumtry.com/guides/canadian-refund-rules-for-merchants.md): What Quebec and Ontario require for online sales, why a posted refund policy can create legal exposure, and how to set these rules as your own policy in Lumtry. - [Pricing](https://lumtry.com/pricing.md): Lumtry plans by case volume and seats, from Free to Enterprise, with add-ons and capped outcome fees. Start free or hear when billing opens. - [MCP server](https://lumtry.com/developers/mcp.md): Connect an AI assistant or your own agent to your refund workspace through the Model Context Protocol. Read cases and propose actions, never move money. - [Trust](https://lumtry.com/trust.md): How Lumtry secures merchant data, keeps AI inside a deterministic refund policy, lists every sub-processor and explains data flow and retention. Review it. - [llms.txt](https://lumtry.com/llms.txt)