Use case

Questions about your data

Ask a question in plain words and get the answer from real rows, with the query that produced it. Writes are refused.

Questions about your data: the workflow in the SynapseOrch editorQuestions about your data: the workflow in the SynapseOrch editor

The workflow in the editor, captured from the open-source app.

Steps
8
Agents
1
Where a person answers
1
Models
Yours to choose, per step: a small one for sorting and checks, a strong one where judgment matters
Stops at
Its turn and time limits; on Cloud, also your plan’s limit

The workflow

What it solves, and how it runs.

The problem

Text-to-SQL looks great in a demo and goes wrong quietly. The fix is checking the query before it runs, refusing anything that writes, and showing the query beside the answer, so a person can see what was asked of the data.

Why a fixed path matters here

The connection is a stored credential and writes stay off unless you turn them on, so the blast radius is a decision you made, not one the model made.

How it runs

  1. 1The tables in your database are listed
  2. 2A model reads the schema of the tables the question needs and drafts one query
  3. 3A second step checks it is one read-only SELECT with a limit, and sends it back if not
  4. 4The query runs with writes refused, as the user you connected
  5. 5The answer comes back in plain words with the SQL; if no rows match, it asks you to narrow the question

Run it

On our cloud, or on your servers.

On SynapseOrch Cloud

Describe it to Build with AI in your own words. It drafts the steps, the model for each and where you answer, and builds nothing until you approve the plan.

On your own servers

database-question-answering.bundle.json holds the agents and the workflow together. Import it from Settings → Import / Export in a self-hosted install, add your own connections, and run it.

It needs

  • A database connection (PostgreSQL, MySQL or SQLite) in Settings
  • A model for the agents, from any provider you set up

Two ways to run it

Let us run it, with nothing to install and no API key needed. Or run it on your own servers, free and open source.

Free plan, no card needed.