Closing the Loop — What Happens After a Dispute is Decided

Release v0.1.323

Deposit disputes don't end when the adjudicator delivers their verdict. For agents, a ruling is also a learning opportunity — a detailed signal about which deductions hold up under scrutiny and which don't. Until now, extracting that signal meant reading through adjudicator reports manually. Starting with v0.1.323, the platform does that work for you.

From outcome to insight, automatically

When an adjudication outcome is recorded, the platform immediately sends the responsible agent a structured lessons-learned summary — both as an in-app notification and by email. The summary breaks down:

• Which deduction categories were awarded — where the adjudicator agreed with the claim.
• Which were rejected — and the structured reasoning tags explaining why.
• The award ratio — a clear figure showing what proportion of the total claimed amount was granted.

This isn't a raw copy of the adjudicator's report. It's a structured, scannable digest designed to be acted on: agents can see at a glance whether their cleaning claims were upheld, whether their damage evidence was considered sufficient, and where future claims in similar cases might need stronger support.

Teaching the AI from real outcomes

The same event that triggers the agent notification — dispute_outcome.recorded — also fires an ai.feedback.dispute_outcome event that flows directly into the platform's AI assessment system.

Every time a dispute is resolved, the AI learns:

• Which types of deduction, in which contexts, adjudicators are most likely to uphold.
• Where the gap is between what an AI assessment predicted and what the adjudicator actually decided.
• How to weight future recommendations to better reflect the standards applied at adjudication.

Over time, this feedback loop means the AI's suggestions get progressively closer to what a human adjudicator would conclude — reducing the number of avoidable disputes and improving outcomes for landlords, agents, and tenants alike.

Why this matters

The goal of this platform has always been to prevent disputes before they start, and to resolve them fairly and quickly when they do arise. Post-adjudication feedback is a critical part of that cycle. Without it, agents repeat the same mistakes, and AI assessments remain static.

With v0.1.323, every resolved dispute becomes a signal that makes the next one less likely.