BlazeSQL
BlazeSQL is an AI data analyst that turns plain English into SQL across twelve databases and warehouses, grounds answers in a knowledge base of business rules, and runs scheduled agent reports to email, Slack and Teams, with a desktop app that keeps results local.
BlazeSQL, from Blaze Analytics vGmbH in Luxembourg, is an AI data analyst that converts plain English questions into SQL and returns answers, graphs, dashboards and reports. It connects to Snowflake, BigQuery, SQL Server, PostgreSQL, MySQL, MariaDB, Oracle, Redshift, Databricks, Amazon Athena, ClickHouse and SAP SQL Anywhere, and offers separate modes for technical and non technical users.
Answers are grounded in a Knowledge Base of notes on definitions, metrics, business rules and instructions kept at database, schema, table or column level; Blaze retrieves the notes that apply to each question, proposes new notes from user corrections for an admin to approve, and checks itself against a set of Training Questions the customer maintains. Agent Reports run on a daily, weekly or monthly schedule, alert on significant changes and deliver to email, Slack or Microsoft Teams. Custom functions and Dynamic Integrations let the chatbot call external APIs and take actions in other services, with an optional approval step before a custom function runs.
A Query Agent API and an MCP server with twelve tools are available on Team Advanced, along with white labeling and embedding. Enterprise adds SSO over SAML or OpenID Connect, access groups restricting tables and columns, row level controls and self hosted storage for query results. Model calls run on Google Vertex AI with zero data retention, the service is hosted on Google Cloud in EU and US regions with specific regions on request, and the desktop app's offline mode keeps results on the user's device. BlazeSQL relies on Google Cloud's SOC reports and signs HIPAA BAAs on enterprise contracts; it states a full on premise deployment is not offered.
Vendor details
Canonical URL
https://blazesql.com
Category
Data analyst agent
Company status
independent
Use cases & customers
Integrations
Databases and warehouses: Snowflake, BigQuery, SQL Server, PostgreSQL, MySQL, MariaDB, Oracle, Redshift, Databricks, Amazon Athena, ClickHouse, SAP SQL Anywhere. Delivery to email, Slack and Microsoft Teams; Claude connector and ChatGPT integration; custom functions and Dynamic Integrations call external APIs; Query Agent API and MCP server; white label and embedding.
In practice
A non technical operations lead asks about last month's delivery performance in plain English, and BlazeSQL writes the SQL using the Knowledge Base's definitions, runs it and adds the graph to a dashboard.
A finance team schedules a weekly Agent Report that analyzes revenue in the background and posts to Slack only when something changed significantly, with an explanation of what happened and what to do.
A software vendor on Team Advanced embeds BlazeSQL in its own product under its own brand and uses the Query Agent API and MCP server to let other tools ask questions of the same data.
Agentic Index coverage score
9.0 / 14 capabilities · 64%
| Integrations & Tool Calling | Full |
|---|---|
|
Beyond twelve databases and warehouses, custom functions connect the chatbot to external APIs with bearer auth so it can send emails, create documents or schedule events from natural language, and Dynamic Integrations let it take actions in any service with an API when given a read write key. Sourcehelp.blazesql.com/en/articles/13585559-custom-functionsread 2026-09-28 |
|
| Workflow Orchestration | Partial |
|
One question can run research, database queries and PDF and interactive report generation, and Agent Reports analyze data in the background, but these are the product's own sequences; no branching, conditions, multiple agents or flow the buyer configures is documented. Sourceblazesql.comread 2026-09-28 |
|
| Knowledge Grounding & RAG | Full |
|
A Knowledge Base of Knowledge Notes (definitions, metrics and business rules, instructions, clarification prompts) is kept at database, schema, table or column level, and Blaze filters and retrieves the notes that apply to each question before answering; notes are added by users or proposed from corrections and approved by an admin, with no retraining. Sourcehelp.blazesql.com/en/articles/13585549-help-your-ai-assistant-learnread 2026-09-28 |
|
| Human Oversight & Guardrails | Full |
|
A custom function can be set to require approval before it executes, so a person confirms the external action before it commits, and knowledge notes the AI proposes from corrections wait for admin approval; access groups restrict tables and columns, and the MCP server runs read only SQL. Sourcehelp.blazesql.com/en/articles/13585559-custom-functionsread 2026-09-28 |
|
| Security, Identity & Governance | Partial |
|
Roles, access groups restricting tables and columns, row level access controls and SSO over SAML or OpenID Connect on Enterprise are documented. For compliance, the Compliance Guide cites Google Cloud's SOC 1, 2 and 3 reports and offers a HIPAA BAA, and names no attestation of BlazeSQL's own. Sourcehelp.blazesql.com/en/articles/13673902-compliance-guideread 2026-09-28 |
|
| Observability & Auditability | Partial |
|
Answers carry the generated SQL (technical mode, and the query field in the API response), and chats, queries and results are retained until deleted, but no admin view of the agent's steps, audit log or trace export is documented. Sourcehelp.blazesql.com/en/articles/13585561-natural-language-query-apiread 2026-09-28 |
|
| Memory & State Persistence | Partial |
|
Chats, queries, dashboards and results are kept until the user deletes them and a conversation carries its context, but the learned corrections become Knowledge Notes in the knowledge base, and no separate memory the agent keeps with a stated scope and lifetime is described. Sourceblazesql.com/privacyread 2026-09-28 |
|
| Deployment & Data Residency | Full |
|
BlazeSQL runs on Google Cloud in EU and US regions and Enterprise customers can request specific GCP regions for residency; the desktop app's offline mode keeps query results on the user's device, and Enterprise adds self hosted storage for query results. The vendor states a fully on premise deployment is not offered. Sourceblazesql.com/privacyread 2026-09-28 |
|
| Prebuilt Agents, Templates & Packs | Not documented |
|
Customers build their own dashboards, agent reports and knowledge notes; no prebuilt agents, report templates or packs are offered. Sourceblazesql.comread 2026-09-28 |
|
| Triggers & Channel Coverage | Full |
|
Agent Reports run on a daily, weekly or monthly schedule the customer sets, analyze the data in the background, alert on significant changes and deliver by email, Slack or Microsoft Teams, with a significance filter that holds back reports when little changed. Sourcehelp.blazesql.com/en/articles/14175981-agent-reportsread 2026-09-28 |
|
| Model Flexibility & Routing | Not documented |
|
Model calls go to Google Vertex AI with zero data retention, and the Team Advanced plan adds an unnamed Advanced AI Model as a tier; the customer does not choose providers or models. Sourceblazesql.com/privacyread 2026-09-28 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
The Query Agent API (POST api.blazesql.com/natural_language_query_api, API key) generates and runs SQL and returns the query, results and explanation, and the MCP server at mcp.blazesql.com (OAuth 2.0 with dynamic client registration) exposes twelve tools including execute_sql, read_dashboard, graph_sql_result and add_graph_to_dashboard, which saves graphs to dashboards; both on Team Advanced. Sourcehelp.blazesql.com/en/articles/13901955-blazesql-mcp-server-documentationread 2026-09-28 |
|
| Testing, Debugging & Optimization | Full |
|
Training Questions are a set of representative business questions the customer keeps and checks Blaze's SQL and results against, the stated way to measure accuracy before onboarding non technical teams, and agent reports can be test run and previewed before they are activated; no numeric score is described. Sourcehelp.blazesql.com/en/articles/13585549-help-your-ai-assistant-learnread 2026-09-28 |
|
| Browser & Computer Use | Not documented |
|
The Team Advanced plan names a Computer Agent for deep research and PDF generation and the homepage shows it researching outside benchmarks, but no browser, desktop or computer control, or how it reaches the web, is documented. Sourceblazesql.com/team-pricingread 2026-09-28 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Pricing
From $150 per month (Individual Pro); Team from $400 per month for 3 users
per month (Individual); per month for 3 users plus per additional user (Team)
What is public
Every Individual and Team plan price with its inclusions and per user add on; Enterprise is custom.
Variable cost rationale
Plans are flat monthly fees with unlimited requests; cost grows only with added users.
Overage / add-ons
Team plans add $50 or $75 per user past the three included; requests are unlimited
Sales call required
Mixed (some tiers require a call)
Free / trial
14 day free trial on Team; no free plan
Lowest paid plan
Individual Pro ($150 per month)
Related vendors
- Actian AI Analyst — Wobby, now sold as Actian AI Analyst, answers plain language…
- AnswerRocket — Enterprise AI agent platform for analytics built around Max, which…
- AskEarth — Geospatial intelligence agent (GIA) that answers plain language…
- AskYourDatabase — AskYourDatabase is a conversational data analyst that turns plain…
- Athenic AI — Athenic AI is a data analyst agent that connects databases,…
- Brewit — Brewit is a conversational data analyst that writes SQL and…
Alternatives to BlazeSQL
The closest documented capability profiles to BlazeSQL among data analyst agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Actian AI Analyst9.5 / 14Adds documented Prebuilt Agents, Templates & Packs and Model Flexibility & Routing
- Brewit6.5 / 14A lighter documented profile than BlazeSQL
- Defog.ai7.5 / 14Fuller documented coverage on Security, Identity & Governance
- Chord8.0 / 14Adds documented Prebuilt Agents, Templates & Packs
- Opensee10.0 / 14Adds documented Prebuilt Agents, Templates & Packs and Model Flexibility & Routing
- Querio9.0 / 14Adds documented Prebuilt Agents, Templates & Packs
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded