Overview

​

One Control Pane, Any Agent Builder

​

The Problem with Traditional Approaches

• ❌ API Development: Building custom APIs for each agent builder
• ❌ Scattered SQL: SQL queries scattered across different systems
• ❌ No Central Control: Can’t update queries from one place
• ❌ Complex Integration: Different integration patterns for each builder
• ❌ Maintenance Nightmare: Changes require updates everywhere
• ❌ No Visibility: Hard to see how agents are using data

​

The Pylar Solution

• ✅ One MCP Server: Single connection point for all agent builders
• ✅ Centralized Views: Define data access once in Pylar
• ✅ One Control Pane: Update views and tools from Pylar, changes reflect everywhere
• ✅ No API Code: No need to build or maintain APIs
• ✅ No SQL Changes Everywhere: Update SQL in one place, all agents get updates
• ✅ Full Visibility: See how all agents interact with your data through Evals

​

How It Works

​

Step 1: Define Views in Pylar

• Write SQL queries
• Join across multiple databases
• Define exactly what data agents can access

​

Step 2: Build MCP Tools

• Use AI to generate tools from natural language
• Or configure tools manually
• Test before publishing

​

Step 3: Publish and Get Credentials

• MCP HTTP Stream URL: Connection endpoint
• Authorization Bearer Token: Secure authentication

​

Step 4: Connect Any Agent Builder

• Code-driven (LangGraph, OpenAI Agents, custom code)
• No-code (Zapier, Make, n8n)
• AI-native (Claude Desktop, Cursor, Windsurf)

​

Step 5: Control from Pylar

• Changes reflect automatically in all connected builders
• No need to update code or configurations elsewhere
• Monitor everything through Evals

​

Universal Compatibility

​

Code-Driven Agent Builders

• LangGraph: Build stateful agent workflows
• OpenAI Assistants API: Create AI assistants
• Custom Python/JavaScript: Any MCP-compatible code
• LangChain: Framework for LLM applications

​

No-Code Agent Builders

• Zapier: Connect apps and automate workflows
• Make (Integromat): Visual automation platform
• n8n: Open-source workflow automation
• Any HTTP-based tool: Tools that support HTTP requests

​

AI-Native Builders

• Claude Desktop: Anthropic’s desktop AI assistant
• Cursor: AI-powered code editor
• Windsurf: AI coding assistant
• Any MCP-compatible tool: Tools supporting MCP protocol

​

Key Benefits

​

For Developers

• No API Development: Skip building custom APIs
• Standard Protocol: Use MCP, the emerging standard
• Focus on Logic: Build agent logic, not data access code
• Easy Updates: Change data access without touching agent code

​

For Non-Developers

• No Code Required: Connect via HTTP requests
• Visual Interfaces: Use familiar no-code tools
• Easy Setup: Just paste credentials
• Automatic Updates: Changes happen automatically

​

For Organizations

• Centralized Control: One place to manage all data access
• Security: Views are the only access level
• Compliance: Full audit trail of all access
• Scalability: Add new agent builders without complexity

​

The Workflow

┌─────────────────┐
│   Your Views    │  ← Define once in Pylar
│   (Pylar IDE)   │
└────────┬────────┘
         │
         ↓
┌─────────────────┐
│   MCP Tools     │  ← Create tools on views
│   (Pylar)       │
└────────┬────────┘
         │
         ↓
┌─────────────────┐
│  Publish Tools  │  ← Get credentials
│  (Pylar)        │
└────────┬────────┘
         │
         ├─────────────────┬─────────────────┬───────────────┐
         ↓                 ↓                 ↓               ↓
┌──────────────┐  ┌──────────────┐  ┌──────────────┐  ┌──────────────┐
│   LangGraph  │  │    Zapier    │  │   Claude     │  │   Custom     │
│   (Code)     │  │   (No-code)  │  │   (AI)       │  │   (Any)      │
└──────────────┘  └──────────────┘  └──────────────┘  └──────────────┘
         │                 │                 │               │
         └─────────────────┴─────────────────┴───────────────┘
                              ↓
                    ┌─────────────────┐
                    │   Evals (Pylar) │  ← Monitor all from one place
                    └─────────────────┘

​

Control and Iteration

​

Update Views

• All connected agents automatically get updated data
• No code changes needed in agent builders
• No configuration updates required

​

Optimize Tools

• See how agents actually use your tools
• Identify optimization opportunities
• Update tools and see improvements immediately

​

Iterate Quickly

• Test changes in Pylar
• Publish updates
• All agents get updates automatically
• Monitor impact through Evals

​

Next Steps

• Claude Desktop - Connect to Claude Desktop
• Cursor - Connect to Cursor
• LangGraph - Connect to LangGraph
• Zapier - Connect to Zapier
• Make - Connect to Make
• n8n - Connect to n8n
• OpenAI Agent Builder - Connect to OpenAI