Learn how to create a match rule with the AI agent, review its test results, and deploy it to your Reltio tenant through the integrated development experience.

1. In Cursor or VS Code, open the AI agent.
2. Describe the match rule in a natural language prompt.

   Include the following details:

   Detail	What to specify
   Entity type	The entity type that the rule applies to, for example Individual.
   Attributes	The attributes to compare, and whether each attribute matches exactly or fuzzy.
   Outcome	Whether matches go to review or merge automatically.
   Tenant ID	The tenant that the rule is for, if several tenants are connected to your workspace.
   Filters or sources	Optional. A condition that limits the rule, for example a type of HCP, or the sources whose records the rule compares.

   The following prompts are examples:

   • Create a suspect match rule for Individual on fuzzy first name, exact last name, and exact email.

   • Match Organizations on fuzzy name and exact city or ZIP code.

   • Never merge two Organizations with different tax IDs automatically.

3. Answer the questions of the AI agent.

   The AI agent asks only about details that it cannot find in your prompt or in your business configuration, such as an attribute name that it cannot find. Unless you specify otherwise, the AI agent uses the following defaults:

   Setting	Default
   Rule type	Suspect, so matches go to review. If your prompt mentions match scores, the AI agent builds a relevance-based rule instead.
   Values compared	Operational values only, which are the values that win under your survivorship rules.
   Comparator and Token Classes	The comparator/token classes that Reltio recommends for the kind of attribute, for example DoubleMetaphoneMatchToken and DoubleMetaphoneComparator for Individual names.
   First-name synonyms	Turned on for approximate first-name matching, so that Bob matches Robert.

4. Review the rule and the test results that the AI agent shows.

   The AI agent shows a description of the rule, the test records with the expected and actual result for each pair, and any notes, for example that a rule on names alone might merge different people. The heading of the results states how the AI agent obtained them:

   Status	Meaning
   Verified against environment/tenant	Your connected Reltio tenant ran the tests. The results show how Reltio matches the records.
   Not verified against a tenant	The tests did not run against a Reltio tenant. The AI agent shows the results that it expects and states the reason.

5. If the results are not what you expect, tell the AI agent what to change.

   The AI agent adjusts the rule and tests it again. When the results are what you expect, the AI agent adds the rule to the entity type in Business Config/BusinessConfig.json.

6. Open Business Config/BusinessConfig.json and review the new rule in the match rules of the entity type.

   To change the rule, for example its scope or its type, edit it in the file or ask the AI agent to change it.

   

7. Deploy the rule. Do one of the following:

   • To deploy the rule, in the business configuration tree of the extension view, right-click the tenant and select Apply Configuration to Tenant.

   • To deploy the rule through Git, commit the change on a feature branch, and then promote the change.