Task resolution is an agent guardrail. It evaluates how well your agents interpret the user's intent, as expressed in a given conversation, and how well it resolves the task.
Required Editions
| Available in: This change applies to Lightning Experience in Enterprise, Performance, Unlimited, and Developer editions with Foundations or Agentforce 1 editions. Setup for AI agents is available on the desktop site. |
Task Resolution measures how effectively the agent understands and resolves the user’s task over the course of a multi-turn conversation. The metric evaluates whether the agent delivered a meaningful outcome that satisfies the user’s intent expressed in the conversation.
Task Resolution Scores
The metric produces structured, non-binary outputs for each conversation. Scores reflect how well the agent understood and handled the task, completed the work, and addressed the query.
| Score | Description |
|---|---|
| Fully Resolved | Indicates that the agent handled all parts of the task. |
| Partially Resolved | Indicates that the agent handled some tasks but gaps remain. |
| Unresolved | Indicates that the agent didn't handle the tasks or address the query. |
Explanations for Outcomes That Aren't Fully Resolved
Also, task resolution captures reasoning for any outcome that isn't Fully Resolved. This reasoning explains why the conversation failed to reach full resolution.
Scores and Explanations Logged in the Generative AI Audit Trail
Logs for Task Resolution scores and explanations are in the generative AI audit trail in a data model object (DMO) in Data 360. Use this information to see how well your agent is interpreting user intent and delivering on what's requested. The data can also help you determine whether the subagent instructions need improvement or if more instructions are needed.
Monitor and Track Task Resolution
Get insights from the collected data to spot trends, interpret results, and remedy agent issues.
- Build your own Data 360 dashboards and reports. To learn more, including required permissions, see Data 360 Reports and Dashboards.
- Run queries.
Here are relevant attributes associated with Task Resolution in the Einstein Generative AI Content Quality DMO in the Data Model for Generative AI Audit and Feedback.
| Attribute | Description | Example |
|---|---|---|
id |
Unique event ID | |
parent |
Generation ID (links to the agent response) | |
detectorType |
Metric name | TaskResolution |
category |
Evaluation result | FULLY_RESOLVED, PARTIALLY_RESOLVED,
UNRESOLVED
|
value |
Explanation text | User intent was to...
|
orgId |
Customer org ID | |
cloud |
Cloud type | einstein
|
timestamp |
Event timestamp |
- Create Data 360 Reports to Monitor Task Resolution
Create Data 360 reports to monitor how well agents carry out a user’s tasks to completion in a session.