How Does Wren AI's Semantic Layer Improve SQL Generation Accuracy?

Wren AI uses a Model Definition Language (MDL) that encodes your business metrics, joins, and definitions. Instead of guessing from raw DDL, the LLM references this semantic context to produce SQL that reflects your actual business logic, reducing incorrect joins and metric misinterpretation.

What LLM Providers Are Compatible With Wren AI?

Wren AI supports OpenAI, Anthropic, DeepSeek, Groq, Azure OpenAI, Google AI Studio (Gemini), Vertex AI, AWS Bedrock, Databricks, and self-hosted models through Ollama. Frontier models like GPT-4o and Claude Sonnet deliver the best accuracy.

Can Wren AI Be Self-Hosted On Private Infrastructure?

Yes. Wren AI offers a fully open-source edition deployable via Docker on your own servers. It supports air-gapped environments where no data leaves your network, making it suitable for regulated industries with strict data residency requirements.

How Does The Embedded AI API Work For SaaS Products?

The Embedded AI API allows developers to integrate Wren AI's text-to-SQL, chart generation, and summary endpoints directly into their applications. Token-based pricing applies only when an AI model processes a request. Non-AI calls like metadata lookups do not consume credits.

What Databases And Data Warehouses Does Wren AI Connect To?

Wren AI natively connects to BigQuery, Snowflake, Redshift, Databricks, PostgreSQL, MySQL, Microsoft SQL Server, ClickHouse, Oracle, Trino, Athena, and DuckDB. CSV file uploads are also supported, and new connectors are prioritized based on community votes.

Wren AI

Name: Wren AI
Availability: InStock
Author: Canner, Inc.

by Canner, Inc. • Palo Alto, California, USA (HQ) • Founded 2018

Open-Source Generative BI With Text-to-SQL Semantic Intelligence

No reviews yet

6 3

text-to-sql generative-bi semantic-layer open-source data-analytics

Visit Wren AI

Follow:

Trust Score

Based on ratings & reviews

4.2 /5

21 reviews

Views

Clicks

Pricing

From $179/mo

What is Wren AI?

Wren AI is an open-source Generative Business Intelligence platform built by Canner that converts natural language questions into governed SQL queries, charts, and actionable insights. Powered by a semantic layer (MDL) that grounds LLM outputs in verified business definitions, it eliminates SQL guesswork and ensures query accuracy across 15+ data sources including BigQuery, Snowflake, PostgreSQL, and MySQL.

Wren AI supports multi-LLM configurations with OpenAI, Anthropic, and self-hosted models via Ollama. Teams can embed conversational analytics into their own products through its API, build real-time dashboards, and manage a centralized knowledge base — all while maintaining row-level and column-level security controls.

Wren AI Demo Video

Key Features 8

Open-Source Text-to-SQL GenBI Platform With 15K+ GitHub Stars

AI-Powered Semantic Layer That Grounds LLM Outputs in Business Definitions

Natural Language Conversational Analytics With Explainable SQL Generation

Auto-Generated Charts, Dashboards, and AI-Written Insight Summaries

Connects to BigQuery, Snowflake, PostgreSQL, MySQL, and 15+ Data Sources

LLM-Agnostic Architecture Supporting OpenAI, Anthropic, Gemini, and Self-Hosted Models

Embeddable APIs for Integrating BI Into Slack, Teams, and Custom Apps

Self-Hostable via Docker With Row-Level Security and Role-Based Access Controls

Who Is Wren AI For

1 Who Is Wren AI For: Data Analysts Querying Databases Without SQL

2 Business Intelligence Teams Building Self-Service Dashboards

3 SaaS Product Managers Embedding Conversational Analytics Via API

4 Data Engineers Standardizing Metrics Through A Semantic Layer

5 CTOs Evaluating Open-Source BI For On-Premises Deployment

6 Marketing Analysts Generating Ad-Hoc Reports From Multi-Source Data

Pros & Cons

Pros 4 benefits

Fully Open-Source Core Engine
15+ Native Data Connectors
SOC 2 Type II Certified
Self-Hosted Air-Gapped Deployment

Cons 3 limitations

Credit-Based Consumption Adds Complexity
Accuracy Depends On Model Choice
Limited Free Tier Capacity

Frequently Asked Questions

5 questions

What's New

Slack Interactive Mode And Embedded Threads 0.52.0

Added Slack Interactive Mode for real-time conversations, Thread Tracing for feedback-driven quality iteration, and Embedded Threads component for integrating AI conversations into external products.

Mar 16

Multi-Tab Projects And New Connectors 0.51.0

Introduced multi-tab support for different project usage, added Athena Spark and Spark connectors, implemented thread tracing for user feedback loops, and increased default model max count to 200.

Feb 13

View all updates

User Base

10K+ users

Active Users

Security & Privacy

Customer-controlled (self-hosted) or Cloud

SOC 2 Type II

Data encrypted in transit and at rest Row-level data access controls Column-level data access controls Role-based access control (RBAC) Audit logs and activity tracking Read-only data policy (no writes to source databases) +2 more