Live, read-only access to what you already run — nothing copied into Tamats.
Tamats searches an organisation's connected S3 documents live and keyword-matches against them with nothing copied into Tamats' own storage, answers questions against a connected research database through a guarded text-to-SQL layer that only ever generates a single read-only SELECT statement, and connects to a Shopify or Cafe24 storefront to read the merchant's own live catalogue and orders. Every one of these is read-only and zero-store: disconnect the source and Tamats has nothing left of it, because nothing was ever kept. An organisation with nothing connected gets an honest 'no source connected' rather than a demo dataset standing in for one.
Most research tools ask you to import your data before they'll answer a question about it, which means a second copy of everything sitting somewhere new to secure. Tamats reads the source directly instead — the S3 bucket, the database, the storefront — live, on every query, and keeps nothing of it afterward. The zero-store property isn't a caveat, it's the point: there is no second copy of your data to lose.
What you do with it
Live S3 document search, zero-store
Keyword search runs against the connected S3 source in real time and returns matching object keys with snippets; nothing from the documents is copied into Tamats' own storage at any point.
Guarded, read-only text-to-SQL
A natural-language question against the organisation's own research database generates a single SELECT statement — never an insert, update, delete or any schema change — run through a read-only database role with an enforced row limit, and the generated SQL is shown for transparency.
Shopify and Cafe24 store connections
The Shopify storefront tier needs no credentials at all; an admin-tier OAuth connection and a Cafe24 mall connection both read the merchant's own live catalogue, policies and orders, never an invented one.
Honest absence, not a stand-in dataset
An organisation with no source connected for a given tool gets a plain statement that nothing is connected, never a sample or demo dataset rendered as if it were real.
What it computes over
—The organisation's own connected S3 bucket, read live per query
—The organisation's own research database, queried read-only through a generated SELECT
—The organisation's own connected Shopify storefront or admin account
—The organisation's own connected Cafe24 mall
Where it stops
Zero-store means exactly that — Tamats does not index, cache or persist a copy of any connected S3 document, database row, or storefront record between queries.
The text-to-SQL layer only ever generates a SELECT statement against a read-only role — it cannot write, alter or delete anything in the connected database, by design, not just by convention.
Shopify's admin tier and Cafe24's admin tier both need the organisation to complete an OAuth connection before write-adjacent data (orders, inventory) is reachable — the storefront tiers alone only see what's already public.
Who picks it up from here
The organisation's own engineer or admin, who owns the underlying S3 bucket, database or storefront account being read and manages who else can connect to it.
Questions
Does connecting an S3 source copy my documents into Tamats?
No — search and read run live against the connected bucket on every query, and nothing is indexed, cached or stored afterward. Disconnect the source and there is nothing left behind in Tamats because nothing was ever kept.
Can the database connector write to my database?
No. It only ever generates a single SELECT statement, run through a read-only database role with an enforced row limit — insert, update, delete and any schema change are not possible through this tool, not merely discouraged.
Do I need a developer account to connect Shopify?
Only for the admin tier. The storefront tier needs no credentials at all and reads whatever is already public on the store; the admin tier needs an OAuth connection and unlocks orders and inventory on top of that.