The Only Identifier You Always Needed But Never Had
Cross reference every entity across the enterprise with an ID that never changes, through new data, retraining and migrations.
First run: every resolved entity gets its own ZINGG_ID. Nothing links Tim Chen to T. Chen yet.
Zingg Maintains The Identity Graph For You
Incremental runs resolve only new and updated records, and assign, merge and split clusters automatically, so the ID you stored yesterday still points to the same customer, supplier or product today. Incremental flow
New Records Join Existing IDs
When a new record matches an entity Zingg already knows, it joins that ZINGG_ID, with a cross reference back to its source record.
New Entities Get New IDs
A record with no existing match starts a new entity with a new ZINGG_ID. Nothing else in the graph changes.
Clusters Merge
When a new record matches two clusters, they merge into one entity with a single surviving ZINGG_ID.
Clusters Split
When an updated record no longer matches its cluster, Zingg moves it out, to a better-matching cluster or a ZINGG_ID of its own.
The Same IDs Through Every Migration
Upgrade the model, change the schema or move platforms. Every downstream system still references the same ZINGG_ID it always did.
Mapped Back By Cluster Membership
Zingg maps the clusters from your new run back to your original cluster assignments by cluster similarity, and keeps the original ZINGG_ID wherever a match exists. A new ID is generated only when no match is found.
- Model upgrades
- Infrastructure migration
- Schema evolution
- Data migration
- Platform upgrades
Audit, Compare And Retrieve By ZINGG_ID
ExplainabilityEnterprise
Give it a ZINGG_ID and get back the evidence: the record pairs compared, their similarity scores and how the cluster formed.
ExplainabilityModel DiffEnterprise
Change a rule, a field or a threshold and see exactly which records would move to a different ZINGG_ID before it reaches production.
Model DiffIdentity RAG
Resolve duplicates and name variations before retrieval, so an LLM answers from one complete entity instead of fragments.
Identity RAGZINGG_ID Questions
What is a ZINGG_ID?
A globally unique, persistent identifier for each resolved entity. It stays stable as new and updated records arrive, so downstream systems can rely on it.
Does the ZINGG_ID change when new data arrives?
No. New records that match an existing entity join its ZINGG_ID. Only a new entity with no match gets a new one.
What happens to the ZINGG_ID when clusters merge or split?
Incremental runs assign, merge and split clusters automatically. A new record that matches two clusters merges them into one entity with a single surviving ZINGG_ID. An updated record that no longer fits its cluster moves to a better-matching cluster or gets a ZINGG_ID of its own.
What happens to the IDs when I retrain or change the model?
Run Reassign ZINGG_ID. Zingg maps the clusters from the new model back to your original assignments by cluster similarity and keeps the original ID wherever a match exists. A new ID is generated only when no match is found.
Can I see why records share a ZINGG_ID?
Yes. With Explainability you query a cluster by its ZINGG_ID and see which record pairs were compared, their similarity scores and how transitive matches chained records together.
Which edition has persistent ZINGG_IDs?
Zingg Enterprise. It adds incremental runs with persistent ZINGG_IDs, so new and updated records resolve without reprocessing everything.
One ID For Every Entity
Persistent ZINGG_IDs and incremental runs come with Zingg Enterprise, on your own data platform.