Data Intelligence
Coming Soon

QueryLlama

Ask your database anything. In plain English.

QueryLlama is a natural-language query interface that translates plain English questions into SQL and returns results — all powered by local Llama models running on your own hardware. Analysts without SQL knowledge get instant answers. Engineers get a faster way to explore unfamiliar schemas. No query leaves your server, no credentials leave your network. Connect a Postgres, MySQL, SQLite, or BigQuery source and start asking questions in minutes.

0 bytes
Data leaves your server
Local
Llama inference
SQL
Generated instantly
4+
Database connectors
AUTO ANSWERBOOKING AUTOMATIONVOICEBOT READYMULTI-LANG IVRCRM INTEGRATIONGDPR COMPLIANTCLOUD OR LOCALREST APIAUTO ANSWERBOOKING AUTOMATIONVOICEBOT READYMULTI-LANG IVRCRM INTEGRATIONGDPR COMPLIANTCLOUD OR LOCALREST API

Available on

  • Web
    Any modern browser
Core capabilities

What it does.

01
Natural language to SQL

Type a question. Get a query. See the answer.

QueryLlama sends your question and the database schema to a local Llama model. The model generates a SQL query, the query runs against your database, and you get results in a clean table. The generated SQL is always shown — you can review, edit, and re-run it. Nothing is hidden. Power users can drop into raw SQL mode at any time; the AI layer is always opt-in.

  • Schema-aware prompting — the model knows your table names, columns, and types
  • Generated SQL shown in full before execution
  • One-click edit and re-run on generated queries
  • Multi-step queries with joins detected and explained
  • Result set summarization: 'Here's what this means in plain English'
01
assistant
› How many new users signed up last month?
SELECT COUNT(*) FROM users
WHERE created_at >= date_trunc('month', NOW())
─────────────────────────────────
Result: 1,847 new users in April
02
Multi-source connectors

Postgres. MySQL. SQLite. BigQuery. One interface.

Connect any standard SQL database with a connection string. QueryLlama introspects the schema, indexes table and column names, and builds a schema context that travels with every query to the model. Credentials are stored locally and never sent to any external service. Support for BigQuery enables cost-efficient analytics on large warehouses without exposing credentials or row data.

  • Postgres, MySQL, MariaDB, SQLite at launch
  • BigQuery via service account JSON (no data egress to LLM)
  • Connection string stored encrypted in local keychain
  • Schema introspection with automatic column type annotations
  • Multi-database workspace: switch sources from a sidebar
02
03
Private by architecture

The model runs on your machine. Your data never leaves.

QueryLlama is built around the same data sovereignty principle as AssistantGeneral. The Llama model runs locally via Ollama or a compatible inference endpoint. Only the schema (column names and types) and the natural language question are sent to the model — never the row data. For air-gapped environments, QueryLlama can operate fully offline. No API keys, no cloud calls, no telemetry.

  • Local inference via Ollama (llama3, codellama, mistral)
  • Only schema + question sent to model — row data stays in DB
  • Air-gapped mode: fully offline operation
  • Optional: swap in any OpenAI-compatible inference endpoint
  • No telemetry, no usage analytics, no external calls
03
More capabilities

Everything else it does.

NL to SQL

Llama-powered translation from plain English to executable SQL with schema awareness.

Multi-source

Postgres, MySQL, SQLite, BigQuery. Connection string stays encrypted and local.

Zero data egress

Only table names and column types reach the model. Row data never leaves the DB.

Schema explorer

Visual schema browser with column-level descriptions and sample values.

Result charts

Auto-detect numeric results and render as bar, line, or pie charts in one click.

Query history

Every NL question and generated SQL saved with timestamps. Re-run or fork.

Licensing

Self-hosted with optional managed cloud.

  • Self-hosted (MIT): run on your own server, unlimited queries, unlimited users
  • Managed Starter: €29/mo — hosted inference, 3 data sources, 5 users
  • Managed Pro: €99/mo — 10 sources, 25 users, query sharing, Slack integration
  • Enterprise: custom — SSO, audit log, dedicated inference node
Free self-hosted · From €29/mo managed
See full pricing
Frequently asked

Questions we hear often.

Any model available via Ollama — llama3, llama3.1, codellama, mistral, and more. You can also point it at any OpenAI-compatible local inference server (LM Studio, llama.cpp server, vLLM). Larger models produce better SQL; llama3:8b handles most OLTP schemas well.

Talk to the team that actually builds the software.

Pilot deployments, volume licensing, product demos, security questionnaires — all handled by engineers and product leads, not a routing layer. We respond within one business day.

Schedule a discovery call
Half-hour walkthrough with someone who built the product — no sales script.
Run a pilot deployment
Full-feature evaluation with guided install, configured for your environment.
Email us directly
sales@royalsoftworks.com — we respond within one business day.

Send us a message

Leave your details and we'll follow up within one business day.