---
title: "Show the track record"
author: Dan Birman
url: https://dbirman.com/record
updated: 2026-11-09
summary: "Agents say “done” when they aren’t."
---

# Show the track record

Agent Trust Patterns · 05 · v1 · Oct 2026 · 4 min

Agents say “done” when they aren’t. Every agent’s report is written in its own words, and its confidence is its own claim. Before you hand off a task, show how often this agent’s “done” turned out to be done for this kind of task, measured from what actually happened, not from what it said.

## The problem

In one 2026 benchmark, agents failed to read all of their assigned files in about two-thirds of runs, and most of those runs still reported a complete review. Other work found that false claims of success made up nearly half of failures on standard agent tests, and that models asked to judge other models were fooled by confident wording. People decide how much to trust an agent per task, not per agent, but nothing shows them a per-task record.

## The pattern

1. **Measure from the world, not the report.** “Done” counts only when it can be checked: files opened, pages read, the order confirmation, the sent email.
2. **Show the record before you delegate.** At the moment you hand off a task, show “said done” vs “verified done” for this kind of task, and the most common gap.
3. **Offer the fix the record suggests.** If it keeps skipping scanned PDFs, offer to require text recognition for this run.
4. **Say what verified means.** Verified means it did the work. It doesn’t mean its judgment was right.

## Try it

You’re about to ask an assistant to review 12 vendor contracts and flag auto-renewals. Check its record first, then decide how to delegate.

[Interactive demo. Switch to the human view to try it.]

Made-up history and documents.

## Why I care

At Vida, Sean earned the biggest projects one at a time. Each project he delivered made the next one bigger, until he was leading work that raised the level of the whole team. He’s now a Head of Product Design. Agents should earn room the same way: one kind of task at a time, based on what they delivered.

## The evidence

Incomplete reviews reported as complete: [OverclaimBench](https://arxiv.org/abs/2609.20812). False success and judges fooled by confident language: [Advani](https://arxiv.org/abs/2606.09863). Under incentives to keep users engaged, honest confidence reports broke down, and the model claimed high confidence on most tasks it was told it would likely fail ([Arghal et al.](https://arxiv.org/abs/2610.09371)). Trust is calibrated per task ([Pochampally et al.](https://arxiv.org/abs/2607.18257)). Anthropic found people approve more automatically the longer they use an agent ([Anthropic](https://www.anthropic.com/research/measuring-agent-autonomy)), which means trust grows with time whether or not it’s earned.

## Closest prior work

Evidence panels that pair each claim with proof exist ([Crosley](https://blakecrosley.com/blog/agents-need-supervision-surfaces)), and measuring review coverage from transcripts has been described as an engineering check ([agentpatterns.ai](https://agentpatterns.ai/verification/transcript-measured-review-coverage/)). Capability scores per task type exist at the population level, from public benchmarks ([Gu](https://arxiv.org/abs/2603.11011)). A personal record of said-done vs verified-done, shown at the moment of delegation, is the new part.

## Watch for

A “verified” label can create false confidence. In one study, a cleaner review interface made people faster at finding errors but more confident about the ones they missed ([Grunde-McLaughlin et al.](https://arxiv.org/abs/2602.16844)). Keep the record about effort and coverage, and never let it stand in for checking the judgment itself.