What is Zug Zug?
One record for every spelling your warehouse can't agree on — self-hosted, next to your data.
Every warehouse fills with names nobody agrees on — BCG, B.C.G.,
Boston Consulting Group, all the same company. Zug Zug is where your team pins them
to one approved record, and maintains the reference tables — Country, Currency,
Partner — that everything downstream depends on.
It runs next to your warehouse, reads with a read-only credential, sets up with one command, and it's yours to host.
The enterprise calls this "master data management." The incumbents — Tamr, Stibo, Reltio — cost six figures and are closed. Zug Zug is the small, self-hosted version, for the people who actually know the data.
The shape of it
- Read-only in. Point Zug Zug at a warehouse column with a read-only credential. It scans the distinct values still waiting for a mapping.
- Agreed by a human. The team maps each source value to an approved record in a spreadsheet-fast grid — drafts, review, bulk merge, roles, full audit history. Optional AI suggestions can propose a record, but nothing is saved until a human confirms it.
- Published as plain tables. Publishing materializes
dim_<x>/map_<x>tables you pull into your own warehouse — no lock-in, no proprietary format — that anything reading your warehouse can join directly.
Not entity resolution. Not an app builder.
Curated lists, and the mappings into them — materialized where your warehouse can read them. No automatic merging, no survivorship rules, no app platform. A person decides which values mean the same thing, and the result is a table.