Unify and manage your data

Resolve matches in AgentFlow using the Match Resolver Agent

Learn how to use the Match Resolver Agent to investigate, compare, and resolve potential duplicates in your data with full audit traceability.

Attention: This feature is available to limited users through the Reltio Early Access (EA) program. Interested in finding out more about this feature or participating in our EA program? Get details in topic Early Access (EA) features.

Use the Match Resolver Agent in AgentFlow to resolve duplicate records. You can explore match candidates, compare attributes side by side, and apply merges or rejections directly from the conversation.

Prerequisites
  • You must have read and write access for the relevant entity types.
  • To use data enrichment features, turn on the Web Search toggle in the agent settings.
Follow these steps to resolve matches in AgentFlow using the Match Resolver Agent:
  1. Start a new conversation and select Match Resolver Agent.
    The agent shows a match inventory summary based on your tenant's data.
  2. Enter a prompt like “What’s on my plate?” to view unresolved matches.
    For example, you'll see counts per entity type like 27,001 individuals and 54 organizations with unresolved potential matches.
    Tip: To focus the agent, try a prompt like:

    "I want to focus on extremely high-confidence matches for the entity type Individual."



  3. Pick a record from the list to open a match candidate view.
    The agent shows side-by-side attribute comparisons between the selected profile and its potential match.

  4. Review the recommendation and explanation provided.
    The agent highlights shared identifiers, high name similarity, or suffix patterns (like LLC vs. Ltd) that support the merge suggestion.

  5. (Optional) Turn on web enrichment to validate external attributes.
    The agent pulls from approved sources and displays citations for validation, such as company affiliations or verified addresses.

  6. Confirm your decision by selecting Merge or Mark as Not a Match.
    AgentFlow applies your choice using the MCP server and confirms the result with a link to the updated record.

Result

When you take action, the system records a full audit log. This includes the timestamp, user ID, input, recommendation, tool(s) used, and final outcome.

Best practices
  • Start with specific prompts like “Show matches for high-confidence individuals.”
  • Ask follow-up questions like “Why do these records match?”
  • Verify source citations before merging, especially when enrichment is enabled.
  • If you only need to inspect records, use the Data Explorer Agent instead.
Audit and traceability

AgentFlow logs every tool call and decision in your tenant's Activity Log. You can stream this data to your Security Information and Event Management (SIEM) system for compliance reporting.

  • User ID and timestamp
  • Tool used (for example, merge_entities)
  • Match candidate records
  • Recommendation rationale
  • Final action: merge or reject
To check access requirements and agent availability, see AgentFlow capabilities and permissions.