WrenAI
WrenAI is an artificial intelligence data assistant that integrates and harmonizes information between LLM models and your databases. Its c...
Last verified:
What is WrenAI?
WrenAI is Wren AI is an open-source Generative BI (GenBI) agent that serves as the open context layer for AI agents, enabling data, product, and business teams to explore data through natural language chat, semantic modeling, and a local UI. It transforms raw database structure into a reusable context layer that gives AI agents grounded, governed memory, context, and SQL across 20+ data sources, helping teams build GenBI dashboards and agentic analytics without writing SQL.
The tool includes three core components: Wren UI for connecting data sources, defining relationships, and asking questions; Wren AI Service for retrieval, prompting, SQL generation, and validation; and Wren AI Core (formerly Wren Engine) as the semantic engine that manages metadata and business context through Modeling Definition Language (MDL). Wren AI generates accurate SQL from natural language questions, creates AI-powered charts and visualizations (Text-to-Chart), and produces business intelligence insights in seconds with full auditability.
Wren AI is designed for data teams, analysts, business users, and AI developers who need governed access to company data. It supports enterprises manufacturing, healthcare, finance, retail, and media sectors that require auditable AI analytics. The platform integrates with Slack, Teams, dbt, and provides embedded API access for custom applications, making it suitable for both human users asking questions and autonomous agents requiring semantic understanding of business metrics like revenue, customers, churn, and active accounts.
Key features include semantic modeling with MDL to encode business definitions, row-level and column-level data control for governance, 20+ data source connectors (BigQuery, PostgreSQL, MySQL, Snowflake, DuckDB, ClickHouse, Trino, SQL Server, Databricks, Redshift, Oracle, Athena, Apache Spark), unlimited dashboards and spreadsheets on paid plans, API access for SQL and chart generation, agent skills and memory, and Git-native
WrenAI pricing
Pricing model: Free
Free tier: $0/month billed annually with 80 credits for first 14 days plus 20 monthly free credits (credits expire monthly, do not roll over). Includes basic GenBI features, visualization, knowledge base, 2 projects, 2 members, 10 tables per project, 2 dashboards per project, 3 AI-powered spreadsheets per project. Essential plan: $179/month billed annually with 13,200 annual credits included (credits roll over up to 2x, $0.1 per additional credit). Includes unlimited dashboards, spreadsheets, connectors, unlimited projects, unlimited members, embedded AI API, API management, dbt integration, basic RBAC, permission controls, standard support, SOC 2 Type II, 3 Slack channels. Enterprise plan: $559/month billed annually with 24,000 annual credits included (credits roll over up to 2x, $0.1 per additional credit). Includes all Essential features plus row-level and column-level data control, MCP integration, unlimited integration, advanced RBAC, auditing logs, custom LDAP/AD, custom OIDC, priority support, embedded threads, groups, and advanced permissions. Usage pricing: Ask Question & Follow-up costs 20 credits input + 80 credits output per 1M tokens, Chart costs 20 credits input + 80 credits output per 1M tokens, Spreadsheet costs 1 credit per execution.
WrenAI pros
- Open-source GenBI agent with 15,533 GitHub stars and #1 GenBI ranking
- Supports 20+ data sources including BigQuery, Snowflake, PostgreSQL, MySQL, and DuckDB
- Generates accurate SQL from natural language questions without manual query writing
- Creates AI-powered charts and visualizations through Text-to-Chart functionality
- MDL semantic layer encodes business definitions to prevent LLM hallucinations
- Row-level and column-level data control for enterprise-grade governance
- Audit logs and activity tracking for provable who saw what query produced which number
- Embedded AI API enables integration into custom applications and chatbots
- Git-native semantic models with versioning, branching, PR review, and rollback
- Works with 10+ LLM providers including OpenAI, Azure, DeepSeek, Gemini, Anthropic, Bedrock
- Slack and Teams integration for self-service querying without analyst tickets
- Unlimited dashboards, spreadsheets, and connectors on Essential and Enterprise plans
- dbt Integration for data team workflow compatibility
- Sandboxed multi-step reasoning with traceable and replayable agent steps
- Free open-source OSS edition available for local installation and testing
- 5-minute setup with sample jaffle_shop data to start quickly
- Supports follow-up questions and maintains conversation context
- Export visualizations as SVG, PNG and data as CSV on paid plans
- Multi-tenant cloud, private cloud, and air-gapped on-prem deployment options
- SOC 2 Type II compliance for enterprise security requirements
WrenAI cons
- Free tier limited to 2 projects, 2 members, and 10 tables per project
- Free credits expire monthly and do not roll over to next month
- Embedded AI API and API Management only available on Essential plan and above
- Advanced RBAC, auditing logs, and custom LDAP/AD/OIDC only on Enterprise plan
- Row-level and column-level data control exclusively in Enterprise plan
- MCP Integration limited to Enterprise plan only
- Slack integration on Essential plan limited to 3 channels
- Teams integration on Essential plan is coming soon, not yet available
- CSV download not available on Free tier
- Performance depends significantly on chosen LLM model capability
- Less capable LLM models may cause reduced performance or inaccurate outputs
- General AI GenBI app (Docker-based chat-first v1) is sunset with no security fixes
- Requires selecting powerful LLM models for optimal results, increasing costs
- Free tier has only 20 monthly credits limiting testing of new features
- Advanced semantic modeling UI limited to 10 tables on Free tier
Frequently asked questions about WrenAI
What is Wren AI and what does it do?
Wren AI is the open context layer for AI agents that sits between your data sources and any agent or application needing to query them. It turns raw database structure into a reusable context layer using Modeling Definition Language (MDL) to define relationships, calculations, and business-friendly abstractions. Wren AI helps agents move from simply seeing tables to understanding what a business means by revenue, customer, refund, churn, and active account, generating accurate SQL from natural language questions and creating AI-powered charts and dashboards in seconds.
What data sources does Wren AI support?
Wren AI supports 20+ data sources including BigQuery, PostgreSQL, MySQL, Microsoft SQL Server, ClickHouse, Oracle, Trino, Snowflake, DuckDB, Redshift, Athena, Apache Spark, Databricks, and Amazon SQL Server. The platform is dialect-agnostic and can transpile from Standard ANSI SQL into different SQL dialects, making it compatible with various database engines and data warehouses.
What is MDL and how does it work in Wren AI?
MDL (Modeling Definition Language) is the semantic contract in Wren AI that defines models, relationships, calculated fields, views, and agent-oriented metadata in files you can read, review, version, and fork. You write a small MDL file describing what your data means, and Wren AI plans every modeled query through it. MDL encodes structural context (tables, columns, types, keys), semantic context (business-facing models, reusable calculations), and business context (company definitions like active customer, revenue, churn), preventing LLMs from hallucinating joins and writing wrong SQL.
What is the difference between Wren AI OSS and Wren AI Commercial?
Wren AI OSS is the open-source edition you install on your machine to try with sample data or connect your own data source, including Wren UI, Wren AI Service, and Wren AI Core. It's free with no commitment. Wren AI Commercial is the full Agentic GenBI platform with governance features like embedded API, enterprise support, row-level and column-level data control, advanced RBAC, auditing logs, and custom LDAP/AD/OIDC. Commercial has Essential ($179/month) and Enterprise ($559/month) plans with annual credits included.
How do credits work in Wren AI pricing?
Credits are consumed for AI-powered endpoints. Free tier gets 80 credits for first 14 days plus 20 monthly free credits that expire at end of month and do not roll over. Essential plan includes 13,200 annual credits that roll over up to 2x. Enterprise plan includes 24,000 annual credits that roll over up to 2x. Additional credits cost $0.1 each. Usage pricing: Ask Question & Follow-up costs 20 credits input + 80 credits output per 1M tokens, Chart costs 20 credits input + 80 credits output per 1M tokens, Spreadsheet costs 1 credit per execution. Credits are not consumed for basic non-AI API actions like retrieving metadata.
What API capabilities does Wren AI provide?
Wren AI provides two main API capabilities available on Essential plan and above: SQL Generation converts natural language questions into SQL queries, allowing you to generate SQL, execute queries and get results, and maintain conversation context for follow-up questions. Chart Generation creates beautiful interactive visualizations from query results, generating Vega-Lite chart specifications and creating visualizations from natural language questions. You need an API key from API Management to access these endpoints, and you can monitor usage in API History.
How does Wren AI ensure governance and auditability?
Wren AI is auditable by design as infrastructure, not just workflow. It enforces Row-Level Security (RLS) and Column-Level Security (CLS) at query time across every human or agent. Role-based access control with full activity logs provides provable tracking of who saw what and which query produced which number. Enterprise plan includes advanced RBAC, auditing logs, and custom LDAP/AD/OIDC. The platform supports multi-tenant cloud, private cloud, and air-gapped on-prem deployment, with SOC 2 Type II compliance across all plans.
What LLM models does Wren AI support?
Wren AI supports integration with various Large Language Models including OpenAI Models, Azure OpenAI Models, DeepSeek Models, Google AI Studio Gemini Models, Vertex AI Models (Gemini + Anthropic), Bedrock Models, Anthropic API Models, Groq Models, Ollama Models, and Databricks Models. Performance depends significantly on LLM capability, so Wren AI strongly recommends using the most powerful model available for optimal results, as less capable models may cause reduced performance, slower response times, or inaccurate outputs.
How do I get started with Wren AI?
Getting started is simple: visit the Install guide to set up Wren AI in your local environment within 3 minutes, or start with Wren AI Cloud managed service. You can try sample jaffle_shop data first to learn quickly, then connect your own database. Visit Usage Guides to learn more about features. The free OSS edition lets you install locally, try sample data, connect your data source, and understand how Wren UI, Wren AI Service, and Wren AI Core fit together before upgrading to commercial plans.
What integrations does Wren AI offer beyond databases?
Wren AI offers Slack integration (3 channels on Essential, unlimited on Enterprise), Teams integration (3 channels coming soon on Essential, coming soon on Enterprise), dbt Integration for data team workflows on Essential and Enterprise plans, Embedded AI API for custom application integration, API Management to generate and manage dedicated API tokens for different applications, and MCP (Model Context Protocol) Integration exclusively on Enterprise plan. These integrations enable self-service querying without analyst tickets and custom agent or SaaS feature development.