Evidence Processing with Trustero Intelligence

These playbooks trigger evidence processing agents directly within the Trustero Intelligence interface, enabling you to analyze and transform evidence without leaving the Evidence page.

You can also use these agents as part of control tests. However, running them directly from the Evidence page is useful for ad-hoc analysis and faster iteration.

Available Evidence Processing Playbooks

Cross-reference two evidence sources
- Compare two evidence sources to identify matching, missing, and overlapping rows.
  
  This playbook:
  - Joins two tables into a single dataset
  - Combines columns from both sources
  - Matches rows based on a shared field (for example, email address or unique identifier)
Evaluate tabular evidence
- Analyze tabular evidence against row-level test criteria.
  
  This playbook:
  - Generates a Test Table with row-by-row results
  - Provides summary statistics about the dataset
  - Produces an overall pass/fail outcome

Filter tabular evidence
- Filter rows using natural language criteria.
  
  This playbook:
  - Applies row-level filtering based on your description
  - Creates a new filtered version of the original evidence
Sample tabular evidence
- Generate a randomized subset of your data for analysis.
  
  This playbook:
  - Selects a sub-sample of rows based on specified criteria
  - Creates a new evidence item containing the sampled data

Open the Evidence Processing Assistant

To open the Evidence Processing Assistant, click Ask Trustero Intelligence in the upper-right corner of the Evidence List page.

Trustero Intelligence Evidence Assistant

Run an Evidence Playbook

When you open the assistant, available playbooks are displayed.

To run a playbook:

Select a playbook from the list
Choose the required evidence sources
- Some playbooks require one dataset
- Cross-referencing requires two datasets

If your Evidence List is already filtered, that filtered view is used to simplify selection.

Evidence Selection for Processing

Process Evidence

After selecting your evidence, the agent begins processing the data to generate the result.

Large datasets are supported, and processing is not limited by language model context size.

Processing Summary, View or Save Result

View or Save Results

After processing is complete, you can:

View results directly in the interface
Save the output as a new evidence item

Saved evidence can be attached to controls and used in control testing.

Save as Evidence

Schedule Evidence Playbooks

You can schedule playbooks to run automatically, including alongside automated evidence collection.

This is useful for recurring workflows such as:

Re-running cross-reference comparisons
Applying filters to newly collected evidence

Scheduled Evidence Processing