Skip to main content
Glama

πŸš€ Built an MCP-Powered Multi-Agent Travel Planner

I’ve been exploring how to design practical Agentic AI systems, so I built MCP Trip Orchestrator using Python, Gemini, LangChain, Tavily MCP, and Pydantic.

The project implements 3 core agentic patterns:

πŸ”Ή Supervisor / Orchestrator Coordinates the workflow and manages budget allocation across tasks.

πŸ”Ή Multi-Agent Collaboration Specialized workers independently handle: πŸš† Transport 🏨 Accommodation 🏰 Experiences & Dining

πŸ”Ή Tool Use + Structured Extraction Agents use Tavily MCP for live web search and Gemini + Pydantic to transform unstructured results into structured booking data.

Architecture

User Request
     ↓
Supervisor / Orchestrator
     β”œβ”€β”€ Transport Agent
     β”œβ”€β”€ Stay Agent
     └── Experience Agent
             ↓
        Tavily MCP
             ↓
          Gemini
             ↓
    Structured Output
             ↓
       Final Trip Plan

What I’m focusing on is not just using an LLM, but understanding how to build reliable agentic workflows around LLMs using tools, specialization, orchestration, and structured outputs.

πŸ› οΈ Python | Gemini | LangChain | Tavily MCP | Pydantic

Next step: making the orchestrator fully adaptive so it can re-plan when constraints change or a worker fails.

#AgenticAI #AIAgents #MCP #ModelContextProtocol #GenerativeAI #Gemini #LangChain #Python #MultiAgentSystems #AIEngineering #MachineLearning #OpenToWork

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers