How to Monitor Auto Agent Performance Using the Auto Agent Log in Eve

Last updated: March 10, 2026

Overview

This article shows you how to use the Auto Agent Log to track how your automatic agents are evaluating matters and producing output. The Auto Agent Log is a monitoring tool for Firm Admins that provides full visibility into agent performance after setup, helps you understand why an agent did or didn't run on specific matters, and identifies when trigger conditions need refinement.

Prerequisites

  • You must be a Firm Admin to access the Auto Agent Log

  • At least one Auto Agent must be configured and active in your workspace

Step-by-Step Instructions

  1. Access the Auto Agent Log — Navigate to Agents in the left sidebar, then select the Auto Agent Log tab at the top of the page. You'll see a summary showing how many automatic agents are currently running, followed by cards for each active automatic agent.

  2. Select an Agent to Review — Click any agent card to load its Activity Details below. The Activity Details section shows how the selected agent evaluated every matter in your firm during the selected time window (default: last 30 days). The header confirms which agent you're viewing and the time period being displayed.

  3. Review the Activity Details Table — The table shows four columns for each matter: Matter name, Result, Outputs Generated, and Last Checked. This gives you an at-a-glance view of which matters the agent evaluated and what happened on each one.

  4. Filter by Result Type — Use the four tabs at the top of the table to narrow what you're seeing:

- All — Every matter in the firm, with a total count

- Output generated — Matters where the trigger condition was met and the agent produced work

- Condition not met — Matters where the agent evaluated but determined the trigger condition was not satisfied

- Not yet checked — Matters the agent has not evaluated yet

  1. Understand Why the Agent Produced Output — Filter to Output generated and expand a matter row. Eve surfaces a "Why the agent produced output" explanation that shows exactly what satisfied the trigger condition — including the specific document name, document ID, last updated timestamp, and the reasoning that confirmed the condition was met. This confirms the agent is firing on the right cases and validates that the logic is working as intended.

  2. Understand Why the Agent Didn't Run — Filter to Condition not met and expand a matter. Eve surfaces a "Why the agent did not produce output" explanation that tells you precisely why the trigger evaluated to false on that matter. Common reasons include documents that existed before the trigger was created (so the "new upload" condition wasn't satisfied) or condition logic that may be ambiguous or too narrow for certain matter configurations. Reading these explanations is the fastest way to determine whether your trigger needs modification.

  3. Use the Log to Refine Your Trigger — After an agent has been running for a few days, conduct a review:

- Filter to Output generated and confirm the agent ran on the right matters and the reasoning matches your intent

- Filter to Condition not met and review explanations on matters where you expected output but didn't get it

- If you see a pattern (for example, the agent isn't catching files with certain naming conventions, or it's missing documents that pre-dated setup), go back and edit the agent's trigger condition to address it

- Re-test the updated condition using the trigger testing panel before saving

  1. Interpret 'Not Yet Checked' Status — If you see a large number of matters in the "Not yet checked" state, this is expected behavior, especially shortly after an agent is created. Automatic agents evaluate matters on a rolling basis and do not retroactively process every matter in the firm on day one. Matters will move out of "not yet checked" as the agent encounters updates or scheduled run times going forward.