What is it?

The Resolver agent in Reltio AgentFlow helps you review, enrich, and resolve potential duplicate entities at scale. It works with your existing match configuration to surface potential matches, compare records, recommend actions, and execute merge or “not a match” decisions with full auditability.

Resolver can also use external web research (when enabled) to verify addresses, company information, and other attributes before it recommends an action.

For more information, see Resolver.

Resolve top-priority duplicates

✅ Prompt: We’re experiencing service delays for our top 100 customers because of data quality issues. We need to resolve duplicates for this segment as soon as possible. How can you help?

Why it works (with custom configuration): This prompt gives Resolver a clear business context (service impact), a target segment (“top 100 customers”), and a concrete objective (resolve duplicates). In practice, this pattern assumes the agent has custom instructions that define how to identify your “top 100” customers (for example, a named segment, a saved list, or attribute-based rules). With that in place, the agent can outline an end-to-end plan, identify potential matches in that segment, enrich and compare records, recommend actions, and ask for your confirmation at each step.

✅ Prompt: Get a high-confidence match from our Organization data in New York state for match review. Do some Web Search and confirm if I should merge or not.

Why it works: This prompt explicitly asks the agent to combine internal match scoring with external evidence. Resolver can surface a strong Organization match in the specified region, then use web research (when enabled) to verify key attributes before recommending whether you should merge, keep the records separate, or request more information.

Review potential matches by confidence

✅ Prompt: Show me potential matches for Healthcare Provider profiles with high or strong match confidence.

Why it works: This prompt tells Resolver which entity type to target and which confidence levels to use. The agent can query potential matches using your configured match rules, filter candidates by match score or confidence band, and return a focused list that you can review before moving to detailed comparisons or merges.

View match comparisons

✅ Prompt: Proceed with Solar Turbines.

Why it works: After you select an entity from a segment or list, Resolver retrieves the primary profile and its potential matches, runs web research when enabled, and summarizes each pair side by side. It highlights shared attributes, key differences, match scores, and recommended actions, so you can see why a potential match is considered a candidate for merge or rejection.

Handle attribute-level discrepancies

✅ Prompt: Review the third match with differences and tell me whether I should merge it.

Why it works: This prompt references the specific candidate (the third match) and asks Resolver to focus on differences. The agent can examine address, name, and other sensitive attributes, flag inconsistencies, use external sources to validate questionable values when available, and explain why it recommends merging or keeping the records separate.

Confirm and execute merges

✅ Prompt: Merge all three matches for this customer according to your recommendations.

Why it works: Once you confirm the recommendations, Resolver executes merge operations using Reltio’s standard merge logic. The agent performs pairwise merges until the duplicates are unified into a single golden record, preserves all contributing crosswalks and lineage, and then summarizes the unified profile — including sources, names, phone numbers, and addresses now consolidated into one trusted record.

Best practices