SQLore
FeaturesPricingDownloadAboutContact
Start freeStart for free
SQLore
HomeFeaturesPricingDownloadFAQAboutBlogContact

SQLore © 2026

TermsPrivacyRefunds
Download for macOSDownload for WindowsSign in
Offline-first AI database assistant

Everything It Does
With Your Database

SQLore connects read-only to your own PostgreSQL or MySQL, learns the schema, and answers questions in plain English — on a local model by default, on a live one when you ask for it.

Download SQLoreSee pricing

How It Answers A Question

Three things have to be true before an answer is worth anything: it picked the right tables, it ran somewhere safe, and you can see what it did.

Plans, runs,
and checks itself

It links the right tables, writes the query, reads the result and corrects a bad join before you ever see it.

Plan

Link schema to question

orders · order_items · products selected from 12 tables.

Verify

Re-ran after a bad join

Corrected before the answer was shown.

One engine,
every surface

Desktop app, web server and CLI from a single codebase, over PostgreSQL or MySQL.

Read-only
by construction

Generated SQL is parsed to an AST and validated to one read-only SELECT, then run on a read-only connection.

SQL guard

drop table users
update orders set …
select * from orders

Parsed to an AST — not matched against a blocklist.

It Learns Your Schema
Before It Answers

Tables, columns, keys, distinct values and statistics are introspected once and cached — which is the difference between a plausible query and a correct one.

Custom business rules

Teach it that "active" means status_id = 3 in your database, once.

Value and exemplar indexes

It knows which columns hold which values, so a filter lands on the right one.

Nothing written back

Snapshots, embeddings and history live in our own store. Your database is only read.

orders

Tables12 indexed
Columns164 typed
Relationships23 keys
9 more business rules
Connect now
Capabilities

Everything In The App

Text-to-SQL is the headline. These are the parts that make it usable on a real database.

Chat with your database

Ask in plain English and get an answer, a table of rows, and the query that produced them. Follow-ups keep the context, so "and just the ones from last quarter" works.

AI SQL generation

Agentic rather than one-shot: it links the right tables, writes the SQL, runs it, reads the result and fixes a bad join before you see anything.

SQL explanation

Paste a query someone else wrote and get it back in English — what it joins, what it filters, and which part is doing the work.

Query optimisation

It reads the plan and says what is slow: a missing index, a join in the wrong order, a filter that cannot use one.

Schema intelligence

Tables, columns, keys, distinct values and statistics are introspected and cached, so a filter lands on the column that actually holds the value.

Local RAG over your schema

The retrieval index is built and searched on your machine. A two-thousand-table database still gets the eight tables the question is about.

Offline AI

Run a local model and the entire loop — retrieval, generation, execution — happens on your hardware with no network access at all.

Live AI models, opt-in

Reach for a hosted model on the questions that need one. It is a choice you make per connection, not a default you have to find and turn off.

Read-only by construction

Generated SQL is parsed to an AST and validated to a single read-only SELECT, then run on a read-only connection with a statement timeout.

Secure connections

Credentials stay on your machine and are never sent anywhere. Point it at a read-only role and the guarantee is enforced twice.

Schema refresh

Migrations happen. Refresh a connection and it re-reads the catalog — and nothing is ever written back to your database.

Saved chats and history

Conversations are kept, searchable and re-runnable, so the query you worked out in March is still there in September.

Rolling it out to a team? The server build gives you one shared instance, read-only credentials are supported throughout, and every answer records the query it ran — talk to us about an enterprise deployment, or read the FAQ.

The SQL guardragai/agent/sql_guard.py

“A model that can write SQL
can write DROP TABLE”

1

Statement per query. Anything that is not a single read-only SELECT never reaches your database.

0

Rows sent to a model provider. Your schema is read; your data stays on your machine.

We read your schema, write the SQL, run it read-only, and show you every query.

SQLore platform

Stop Writing The
Same Three Joins

Start free, upgrade when you are asking enough questions for it to matter.

Start for freeContact sales
sqlore.com
Connectionretail_db

staging

analytics

Add connection

MENU

Chat

Reports

Schema

History

Settings

Rows scanned

15,600+18%

Questions

25,800+14%

Avg latency

41 ms−16%

By product

By month

By connection