Relational databases
PostgreSQL, MySQL, and Druid. tableArth.ai reads your tables, joins, and column types, then writes real SQL from your question — the same idea as text-to-SQL, grounded in your schema.
Talk to your database
Ask your database in plain English.
Query external databases such as PostgreSQL, MySQL, MongoDB, and Druid running inside your own private environment — simply by asking. tableArth.ai writes and runs the query for you and returns a grounded answer. No exports, no data leaving your network.
PostgreSQL · production
Ask
Query
You
Monthly recurring revenue by plan for the last 6 months?
MRR grew from $182K to $264K over
six months. Business plan drove most of the
gain, up 61%; Pro up 24%; Starter flat.
MRR by month
USD
FebMarAprMayJunJul
SELECT plan, date_trunc('month', started_at) m,
SUM(amount) FROM subscriptions
GROUP BY plan, m ORDER BY m;
SUM(amount) FROM subscriptions
GROUP BY plan, m ORDER BY m;
Supported databases
Relational and document, both by asking.
Connect the database you already run. tableArth.ai reads its schema, understands the tables or collections, and lets anyone ask it a question in plain English.
PostgreSQL
Database
MySQL
Database
MongoDB
Database
Apache Druid
Database
More connectors
Coming soon
Your data stays yours
It runs where your data lives.
tableArth.ai queries databases inside your own private environment. The point is that nothing has to move: no dumps, no copies of your rows shipped off to a third party to get an answer. You keep control of the connection, the access, and exactly what a model is ever allowed to see.
- No exports — query in place against your database
- Read-oriented — point it at a read replica or read-only role
- Four privacy modes — from full AI to fully local, per source
Most capable
Full AI
Rich answers with chart and insight.
Capability—
Privacy
Masked data
Sensitive values tokenized before the model.
CapabilityPrivacy
Hybrid
Stats only
Model sees stats. Rows stay in your stack.
CapabilityPrivacy
Most private
Local template
Zero external AI calls. Server-side only.
—Privacy
How it works
Question in. Real query out. Grounded answer back.
The answer isn't a guess from a language model — it's the result of a query the engine actually wrote and ran against your database.
Step 01 — Ask
Type your question
Plain English, no schema knowledge. tableArth.ai suggests questions and asks a clarifying one if your request is ambiguous.
No SQL required
Step 02 — Query
It writes the query
Schema-aware SQL for relational databases, or the right query for a document store like MongoDB. It self-corrects a failed query up to three times.
SQL & NoSQL
Step 03 — Answer
Runs it, answers you
The query runs in your environment and the result comes back as a readable answer with the right chart — grounded and checkable.
Answer + chart
SQL and NoSQL
Tables or documents — you just ask.
Document databases
MongoDB. No aggregation pipelines to hand-write — the engine understands collections and nested documents and builds the right query for you. See querying MongoDB in natural language.
FAQ
Querying your databases.
Which databases can I query in natural language?
External databases such as PostgreSQL, MySQL, MongoDB, and Druid. More connectors are being added. Both relational (SQL) and document (NoSQL) databases are supported.
Does my data have to leave my environment?
No. tableArth.ai queries databases running inside your own private environment. The data stays where it lives — there are no exports and no copies of your rows leaving your network to get an answer.
Do I need to write SQL or an aggregation query?
No. You ask in plain English and tableArth.ai writes and runs the query for you — SQL for relational databases, the right query for document databases like MongoDB — then returns the answer with a chart.
How accurate and safe are the answers?
Answers are grounded in a real query run against your data, so they are verifiable rather than guessed. The engine self-corrects a failed query up to three times, and four privacy modes let you control exactly what the model can see.
Can non-technical users use it?
Yes. Anyone who can ask a question can query the database. tableArth.ai suggests relevant questions and asks a clarifying question when a request is ambiguous, so people get to the right answer without knowing the schema.
Get started
Point it at a database. Ask your first question.
Connect a read replica in your environment and let your team query it in plain English — grounded answers, in seconds.