Getting Started with the Data Reconciliation Recipe
The Data Reconciliation recipe compares an asset you want to investigate with a reference asset, row by row, confirms every record lines up, and surfaces the fields where their values disagree. It runs in two phases. Build & Validate configures a Data Diff check between the two assets and activates it. Scan & Reconcile scans every record, lets you decide which side is right for each mismatch, like resolving merge conflicts, and writes the keep and remove sets to your enrichment destination.
Two things must be in place before the recipe can run end to end:
- The two assets must share at least one field with a compatible type, since fields are paired by name and type.
- The datastore of the asset you investigate needs a linked enrichment destination before the Scan step. The recipe blocks in place and offers to link one when it is missing.
The Deep Dive pages explain how the recipe works, what each step does, and how existing checks and anomalies enter the flow. The How-tos walk through each task step by step, and the Troubleshooting, API, and FAQ pages cover the rest.
Permissions
Anyone can open the recipe and browse its steps. Creating the check needs the Drafter team permission on the datastore of the asset you investigate, activating it needs Author, and running the scan and writing the keep set need Editor. Reading the reference asset needs at least Reporter on its datastore. See Permissions for the full matrix.
Deep Dive
Understand what the recipe is, how each phase works, and who can do what.
-
Introduction
What the recipe is, the two phases, and what it produces.
-
How It Works
Left and right, the phases, the prerequisites, what the run reads and writes, and AgentQ's role.
-
Build and Validate
Select Assets, Row Identifiers, Compare Fields, Tolerances, Review, and Validate, step by step.
-
Scan and Reconcile
The full scan, resolving each mismatch, writing the keep and remove sets, and the completion summary.
-
Entry Points and Resuming
Scan in Recipe, Reconcile in Recipe, and how progress is kept between visits.
-
Examples
Real scenarios showing two assets compared, their mismatches resolved, and the keep set written.
-
Best Practices
Guidelines for choosing sides, identifiers, tolerances, and resolutions that hold up.
-
Permissions
The roles and team permissions behind each step of the recipe.
How-tos
Run the recipe and enter it from existing work, step by step.
-
Run the Data Reconciliation Recipe
Go from the Recipes page to a written keep set.
-
Scan a Check in the Recipe
Take an active Data Diff check straight to the Scan step.
-
Reconcile an Anomaly in the Recipe
Open a Data Diff anomaly at the Reconcile step with its mismatches loaded.
-
Link an Enrichment Destination from the Scan Step
Link the enrichment destination the recipe needs without leaving the flow.
-
Resume or Start Over
Continue where you left off, or clear your progress and begin again.
Troubleshooting, API and FAQ
-
Troubleshooting
Known problems while selecting assets, validating, scanning, reconciling, and materializing, and how to resolve them.
-
API
The sequence of calls behind the recipe, and the AI assist endpoints it uses.
-
FAQ
Common questions about the rule type, left and right, tolerances, outputs, AgentQ, and resuming.