Overview of Record Linking Recommendations
Spark AI Record Linking Recommendations leverages generative AI to automatically identify and link data across key GRC elements within your Risk Cloud environment. It streamlines historically complex and labor-intensive control cross-mappings and empowers holistic GRC management by identifying relationships and linking Risks, Controls, Policies, and Incidents.
Key use cases of Spark AI Record Linking Recommendations include:
- Linking internal controls and policies to compliance frameworks
- Linking evidence to controls
- Linking incidents to risks
How to Enable Record Linking Recommendations (RLR) for your Environment
System Administrators can enable Spark AI and its features by navigating to the Risk Cloud Spark AI Card on the Integrations page. This includes a toggle to activate the Record Linking Recommendations (RLR) feature
- Enabling Spark AI RLR at the environment level gives you the flexibility to choose exactly where it applies. While it remains off by default for individual records and steps, you can easily activate it for specific workflows within their Linked Workflow sections (see Configuring Builder Settings for RLR below).
Configuring Builder Settings for RLR
1. By default, RLR is disabled for all your Linked Workflow Sections. It must be configured before it is enabled.
2. To enable RLR, navigate to the Step where your Linked Workflow Setting resides. Make sure a Layout is used. Then under Record Linking & Submission Rules, select "Enable the ability to search for [records] and link them", and click Configure.
- Note: RLR support for Table Reports is coming in Summer 2026.
3. You must select at least one field per workflow. By default, the Primary Field and Summary Fields are pre-selected for you. Spark AI will parse the data from these selections, so we recommend adding any additional descriptive fields that give your records the best context.
4. Once this is configured, you'll now see that Spark AI Recommendations are enabled. You can always disable or edit configurations.
Using Record Linking Recommendations on Records
Once you (the admin user) have enabled and configured Spark AI Record Linking Recommendations, standard users can follow these steps to get started using RLR.
1. Navigate to the record where you want to generate linked record recommendations and have configured a linked workflow section. You will see the Recommendations tab as an option within the linked workflow section. Click on this tab to access the recommendations feature.
2. On the Recommendations tab, click the Generate Recommendations button. Spark AI will automatically provide suggestions based on your current record. At the top of the results, an Analysis Summary shows the total number of matches found, broken down by Strong Matches and Possible Matches, giving you a quick snapshot before diving into the full list.
- Note: If you don't see the Recommendations tab, make sure that a Layout is being used for the Linked Workflow section in the Step Configuration. RLR support for Table Reports is coming in Summer 2026.
3. After reviewing the automated recommendations, you can link any records by clicking the Link button. Then you will see this record in the "View linked records" tab. You can always refresh the record page to re-generate top recommendations at any time.
Frequently Asked Questions
What do the "Strong Match" and "Possible Match" confidence labels mean?
Each recommendation includes a confidence label in addition to a match percentage. A Strong Match (90–100%) indicates a high degree of relevance between the records, while a Possible Match (60–89%) suggests a meaningful but less certain relationship worth reviewing. Records below 60% are not returned.
What is the Match Rationale column?
Match Rationale is an AI-generated explanation describing why a particular record was recommended as a match. It helps you quickly evaluate whether the recommendation is relevant to your use case without having to open each record individually.
How does Spark AI determine match confidence?
Spark AI uses a two-step process: first, a text embedding model pre-filters candidates down to the top 50 most relevant records, then a generative AI model evaluates those candidates to assign a confidence score and generate a rationale. The fields you configure in the Linked Workflow section are what Spark AI reads to make these assessments — so selecting descriptive, context-rich fields leads to better recommendations.
Why don't I see a Recommendations tab on my record?
A couple of things to check:
- Confirm that Spark AI RLR is enabled at the environment level by a System Administrator.
- Verify that RLR is configured on the specific Linked Workflow section in your step's builder settings.
Why am I seeing more (or fewer) recommendations than before?
Previously, Spark AI returned only the top 10 results. With the latest improvements, all Strong and Possible matches are returned, ordered by match strength from highest to lowest. This gives you a more complete picture of potential relationships across your data.