🤖 The Semantic Engine for Model Context Protocol (MCP) clients and AI agents. Provides a semantic layer that connects LLM-powered agents to enterprise data systems with governance and context.
https://github.com/Canner/wren-engineStop fighting with raw database connections in your AI agents. When you're building enterprise AI workflows, connecting to PostgreSQL or BigQuery is just the beginning—the real challenge is getting your AI to understand what "active customer" or "monthly recurring revenue" actually means in your business context.
You've probably built MCP servers that connect to databases before. The connection works fine, your AI can execute SQL queries, but then reality hits:
Raw database access gives you data. Wren Engine gives you business intelligence.
Wren Engine isn't just another database MCP server—it's a semantic layer that sits between your AI agents and enterprise data systems. Think of it as the translation layer that converts business intent into accurate, governed data operations.
Here's what you get:
Business Context Built-In: Define metrics like "active customers" or "monthly churn rate" once, and every AI interaction uses the same calculation. No more inconsistent results across different prompts.
Multi-Source Intelligence: Connect to Snowflake, BigQuery, PostgreSQL, and 8+ other data sources simultaneously. Your AI can join data across systems without you writing complex federation logic.
Governance by Default: Role-based access control and business definitions ensure your AI agents respect the same data governance rules as your BI dashboards.
Instead of this typical MCP database interaction:
-- AI generates this, but is "active" 30 days or 90 days?
SELECT COUNT(*) FROM customers WHERE last_login > NOW() - INTERVAL '30 days'
You get semantic queries that understand your business:
# AI can now ask: "How many active customers do we have?"
# Wren Engine knows your business definition of "active"
result = await wren_query("active_customers", context="current_month")
AI-Powered BI Dashboards: Build agents that generate reports using your exact business definitions. Finance gets the same "revenue" calculation whether they ask the AI or check the dashboard.
Cross-Team Data Agents: Customer success asks about "churn risk" while sales asks about "qualified leads"—both get consistent, accurate data based on your established business rules.
Compliance and Audit Workflows: Ensure AI-generated reports meet the same data governance standards as your official business reporting.
The setup is straightforward if you're already working with enterprise data:
git clone https://github.com/Canner/wren-engine.git
cd wren-engine
./gradlew clean build -x test
java -jar build/libs/wren-engine-*.jar server
warehouse:
type: snowflake # or bigquery, postgres, etc.
url: jdbc:snowflake://your-account.snowflakecomputing.com
user: YOUR_USER
password: YOUR_PASSWORD
Define your business semantics by posting your model definitions to /v1/models
Connect to your MCP client and start querying with business context
Wren Engine slots directly into your existing MCP setup. If you're already using Claude, Cursor, or other MCP clients, you add this as another server in your configuration. The difference is that now your AI understands your business, not just your database schema.
Your development workflow becomes:
As AI agents become more central to business operations, the gap between raw data access and business understanding becomes a critical bottleneck. Wren Engine solves this by bringing enterprise-grade semantic layers into the MCP ecosystem.
You're not just connecting to databases—you're connecting to business intelligence. And that's exactly what enterprise AI agents need to be truly useful.
Ready to try it? Check out the GitHub repository and see how semantic data access changes your AI development workflow.