Back to vendors
D

Defog.ai

Visit site
Entry priceEnterprise (cloud hosted) $5,000 per month; self hosted on annual commitmentFull pricing detail

Defog.ai is an enterprise text to SQL analyst built on its own SQLCoder models. Teams align it to their schema with instructions and golden queries, ask questions in the app, Slack or an API, and can run it on their own infrastructure.

Defog.ai turns plain language questions into SQL for BigQuery, Databricks, MySQL, Postgres, Redshift, Snowflake, SQLite and SQL Server, and returns the SQL alongside tables and charts. It is built on SQLCoder, Defog's own family of open source text to SQL models. Defog also published SQLEval, an open source framework for evaluating generated SQL.

Admins align Defog to their data with written instructions on business rules and golden queries, pairs of questions and verified SQL. Users rate answers, a feedback page collects the results, and an Improve using Feedback feature proposes new instructions from negative feedback. Questions can be asked in the Defog app, in Slack or through an API that carries conversation context. Admin and general user roles separate who can align the model from who can only query, and SSO is available to enterprise customers.

Defog is sold as a cloud hosted Enterprise plan at $5,000 a month or a self hosted Enterprise plan on an annual commitment, deployed as three Docker images with a vLLM model service inside the customer's infrastructure. Defog announced SOC 2 Type II compliance, names no auditor and has published no later attestation.

Vendor details

Canonical URL

https://defog.ai

Category

Data analyst agent

Company status

independent

Use cases & customers

Integrations

Defog connects to BigQuery, Databricks, MySQL, Postgres, Redshift, Snowflake, SQLite and SQL Server, accepts CSV uploads, integrates with Slack and offers a query API (generate_query_chat).

In practice

A regulated enterprise runs Defog's Docker images on its own servers, aligns it with instructions and golden queries for its schema, and lets analysts ask questions while data stays inside its network.

A software team calls Defog's query API from its own product to turn a customer's question into SQL, passing previous context so follow up questions work.

A data lead reviews the feedback page each week, turns good answers into golden queries and uses Improve using Feedback to draft instructions that fix the answers users marked wrong.

Agentic Index coverage score

7.5 / 14 capabilities · 54%

Integrations & Tool Calling Partial

The databases Defog writes SQL for are BigQuery, Databricks, MySQL, Postgres, Redshift, Snowflake, SQLite and SQL Server. It accepts CSV uploads and has a Slack integration. The connectors are databases and a chat surface. Defog names no actions in other systems and no custom tools.

SourceDefog, docs.defog.ai API Access, Setting Up Your Database and Slack Integration pages and homepageread 2026-09-28

Workflow Orchestration Partial

When Defog Agents launched, they handled multi step workflows in SQL, Python and R with a person in the loop. The current product turns questions into SQL queries. Defog names no branching, conditions, multiple agents or flow the customer configures.

SourceDefog, blog post of 5 October 2023 and docs.defog.ai Querying Your Data; defog.ai/blogread 2026-09-28

Knowledge Grounding & RAG Partial

Admins align Defog with schema metadata, written instructions on business rules and golden queries (question and SQL pairs). The backend stores them and builds the prompts sent to the model. Defog names no retrieval over this material and no way to scale it past the prompt.

SourceDefog, docs.defog.ai Aligning Defog To Your Usecase and On Premise Architecture pagesread 2026-09-28

Human Oversight & Guardrails Partial

Only admins may align the model and manage the database, while general users can only query. Admins set the business rule instructions the agent follows, and a person refines golden queries before they are saved. Defog Agents launched with a person in the loop. Defog names no approval step before a query runs.

SourceDefog, docs.defog.ai Managing Users and Aligning pages and blog post of 5 October 2023read 2026-09-28

Security, Identity & Governance Full

Access is split between admin and general user roles, and enterprise customers get SSO. Defog announced SOC 2 Type II compliance, names no auditor and has published no later attestation.

SourceDefog, docs.defog.ai Managing Users page, Pricing page and blog post of 6 October 2023; defog.ai/blog/soc-2-type-iiread 2026-09-28

Observability & Auditability Partial

Users see the SQL behind each answer alongside tables and charts. A feedback page lists the queries users marked right or wrong, with their comments. Defog names no audit log, per query history for admins or trace export.

SourceDefog, homepage and docs.defog.ai Aligning pageread 2026-09-28

Memory & State Persistence Partial

The query API carries previous context through a conversation. Feedback turns into golden queries and instructions that ground later answers. Defog names no memory the agent keeps across sessions.

SourceDefog, docs.defog.ai API Access and Aligning pagesread 2026-09-28

Deployment & Data Residency Full

A self hosted Enterprise plan runs on the customer's own infrastructure. The on premise deployment is three Docker images, including a vLLM model service, with users, metadata and golden queries kept in the customer's Postgres.

SourceDefog, Pricing page and docs.defog.ai On Premise Architecture and Docker Deployment pagesread 2026-09-28

Prebuilt Agents, Templates & Packs Not documented

There are no prebuilt agents, templates or packs. The customer aligns Defog to its own schema with instructions and golden queries.

SourceDefog, docs.defog.ai and homepageread 2026-09-28

Triggers & Channel Coverage Partial

The customer's own users ask questions in the Defog app, in Slack or through the API. Each of these starts with a person. Defog names no schedules, alerts, events, channel for the customer's own customers or embedded widget.

SourceDefog, docs.defog.ai Slack Integration and API Access pagesread 2026-09-28

Model Flexibility & Routing Not documented

Answers come from Defog's own SQLCoder models, served by a vLLM service in the on premise build. Defog names no third party model and no way for the customer to select a provider.

SourceDefog, homepage and docs.defog.ai On Premise Architecture page; defog.airead 2026-09-28

APIs, SDKs & MCP Extensibility Full

An API call (POST api.defog.ai/generate_query_chat with an API key) returns SQL for a natural language question against a named database type, carrying previous conversation context.

SourceDefog, docs.defog.ai API Access pageread 2026-09-28

Testing, Debugging & Optimization Full

Users rate each answer in the product, and a feedback page collects the results. Improve using Feedback proposes new instructions from negative feedback, and good answers become golden queries, so the deployed analyst keeps improving. SQLEval, Defog's open source framework, benchmarks Defog's own models, not the customer's deployment.

SourceDefog, docs.defog.ai Aligning Defog To Your Usecase pageread 2026-09-28

Browser & Computer Use Not documented

The product works through database connections and SQL, and Defog names no browser, desktop or computer control.

SourceDefog, homepage and docs.defog.ai; defog.airead 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

Enterprise (cloud hosted) $5,000 per month; self hosted on annual commitment

per month (cloud hosted); annual commitment (self hosted)

What is public

Defog publishes the cloud hosted Enterprise price; the self hosted plan's price is not published.

Variable cost rationale

A flat monthly fee for the cloud plan and an annual commitment for self hosting; no usage metering is published.

Sales call required

Yes, required for paid access

Free / trial

Not published

Lowest paid plan

Enterprise, cloud hosted ($5,000 per month)

Agentic Index verified 2026-09-28

Alternatives to Defog.ai

The closest documented capability profiles to Defog.ai 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.

  • BlazeSQL9.0 / 14Fuller documented coverage on Integrations & Tool Calling and Knowledge Grounding & RAG
  • Querio9.0 / 14Adds documented Prebuilt Agents, Templates & Packs
  • AskYourDatabase7.5 / 14Adds documented Model Flexibility & Routing
  • Brewit6.5 / 14Fuller documented coverage on Knowledge Grounding & RAG and Triggers & Channel Coverage
  • Powerdrill8.0 / 14Adds documented Prebuilt Agents, Templates & Packs
  • Actian AI Analyst9.5 / 14Adds documented Prebuilt Agents, Templates & Packs and Model Flexibility & Routing

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.