用MongoDB Atlas和LangGraph构建智能场馆运营Agent

Introduction This tutorial starts where most agent demos stop: giving the agent persistent memory, operational context, and a place to write back what happened. An event operator does not just need an agent that can summarize a weather report or generate a generic plan. The operator needs an agent that can remember what happened at prior events, retrieve relevant visitor and venue context, respond to live operational changes, and write the outcome back as memory for the next similar situation. We built this event-venue operator demo with MongoDB Atlas, Voyage AI embeddings, LangGraph, and optional Langfuse tracing. The demo scenario is the MongoDB Open, a fictional premium tennis tournament on Day 6 of play. Rain is approaching, covered hospitality capacity is constrained, and the operator has two different visitor journeys to protect: Mikiko, a first-time attendee trying to make the most of the grounds, and Nina, a premier guest with hospitality expectations and a history the agent can retrieve. This is not a customer case study or a production deployment. It is a fictional builder scenario inspired by real event operations economics. Major tennis events show why these decisions matter: the 2025 US Open broke attendance, viewership, and digital reach records and offered $90 million in total player compensation; USTA has also said the three-week US Open drives more than $1.2 billion in annual economic impact for New York City. Premium fan expectations are high, too: PwC found that 60% of high-income U.S. sports fans would spend more than $250 for a special event, and 20% would spend more than $1,000. Weather adds another layer of risk, which is why the U.S. Census Bureau now tracks the monetary impact of extreme weather on business sales through its Business Trends and Outlook Survey. The MongoDB Open demo agent is not just producing a plausible plan. It reads current venue state, retrieves prior event memory, distinguishes between visitor segments, and acts. At the same time, hospitality capacity is still available, and writes the outcome back so the next disruption can be handled with more context. Check out the full repo here. The demo is split into three layers: A guided, deterministic UI that makes the operator story easy to follow. A hosted Vercel demo that gives readers a public app link. Live API endpoints and scripts for Atlas Vector Search, vector-plus-lexical retrieval, visual-document RAG, LangGraph execution, and optional Langfuse traces, to demonstrate how the stack all works together. What You Will Build By the end of the tutorial, you will have a FastAPI app backed by MongoDB Atlas that can run locally and deploy to Vercel. The app includes: A four-tab guided UI for the event-operations story and live backend validation. Atlas collections for operational state, semantic memory, agent actions, and LangGraph checkpoints. Voyage multimodal embeddings stored in Atlas. Atlas Vector Search for memory retrieval. A hybrid retrieval endpoint that combines vector similarity with lexical scoring. A Vision RAG endpoint that retrieves visual operational documents and passes them to Claude Vision. Optional Langfuse tracing for retrieval calls and the live LangGraph run. A runnable LangGraph script that follows the same rain-delay story. A Vercel deployment configuration for a hosted demo. The current repo should be treated as a reference demo, not a production platform. There is no production auth, no CI suite, and the full LangGraph agent remains a script-based validation path rather than a public hosted endpoint. Architecture Overview The architecture centers on MongoDB Atlas as both the operational and memory layer. Speed matters in the event venue operator scenario because the useful window for action is short. If rain is 20 minutes away and covered hospitality space is filling up, the operator does not need a post-event dashboard or a batch summary a few minutes later. The agent needs to read the current venue state, retrieve relevant memory, decide what to do, and write back the result while there is still capacity to protect the guest experience. That is why the type of database and how it is used are critical system design choices. Operational records, semantic memory, vector embeddings, visual documents, and agent actions all live in the same data layer. The agent does not need to wait for a separate analytics pipeline, sync data into a second vector database, or reconcile what the memory layer says with what the operational system says. Atlas acts as both the system of record and the retrieval layer for the agent loop: perceive what changed, retrieve the right context, take action, and persist what happened for the next event. This is also why the demo keeps memory in MongoDB rather than treating it as a sidecar. The agent is not just retrieving chunks; it is composing operational context. A useful decision may need visitor history, current venue status, hospitality inventory, prior rain-delay patterns, and relevant visual documents at the same time. With Atlas, those pieces can stay queryable together instead of being scattered across separate systems. Caption: MongoDB Atlas stores the demo’s operational state, semantic memory, visual document embeddings, agent actions, and LangGraph checkpoints in one backend. The demo uses four main state layers: Operational records: guests, visits, venue status, weather events, reservations, event metrics, and agent actions. Semantic memory: memory_store, with Voyage embeddings and Atlas Vector Search. Visual documents: operational images embedded into the same memory store as image-derived multimodal embeddings and document metadata. Agent state: LangGraph checkpoints and checkpoint writes. Setup Before you begin, make sure you have: Python 3.12 or later uv installed A MongoDB Atlas cluster with Vector Search enabled (this can be set up for free) An Anthropic API key (or feel free to use an LLM of your choice and reconfigure API keys) A Voyage API key (this can be set up for free) Clone the repo and install dependencies: GitHub repo Copy CodeCopiedUse a different Browsergit clone https://github.com/mongodb-developer/event-venue-operator.git cd event-venue-operator uv sync If you only want to inspect the app before setting up credentials, start with the live Vercel demo. The hosted demo uses the same UI and deployment shape as the repo, while local setup lets you run the full seed, smoke test, Vision RAG, and LangGraph paths yourself. Create your environment file: Copy CodeCopiedUse a different Browsercp .env.example .env Add the required values: Copy CodeCopiedUse a different BrowserMONGODB_URI=mongodb+srv://<user>:<password>@<cluster>.mongodb.net/?retryWrites=true&w=majority MONGODB_APP_NAME=devrel-tutorial-agentic_retrieval-memory-marktechpost MONGODB_DATABASE=event_venue_operator ANTHROPIC_API_KEY=sk-ant-... VOYAGE_API_KEY=pa-... Langfuse is optional for observability: Copy CodeCopiedUse a different BrowserLANGFUSE_PUBLIC_KEY= LANGFUSE_SECRET_KEY= LANGFUSE_HOST=https://cloud.langfuse.com Initialize Atlas: Copy CodeCopiedUse a different Browseruv run python scripts/setup_atlas.py This script creates collections and starts the Atlas Vector Search index, then waits up to 60 seconds for the index to become READY. Then seed text and visual documents: Copy CodeCopiedUse a different Browseruv run python scripts/seed_data.py uv run python scripts/seed_visual_docs.py Start the app: Copy CodeCopiedUse a different Browseruv run python -m event_venue_operator.server Open http://127.0.0.1:8000/. In a second terminal, run the smoke test: Copy CodeCopiedUse a different Browseruv run python scripts/smoke_test.py With the server running in another terminal, the smoke test checks MongoDB health, Atlas Vector Search, hybrid search, visual-document indexing, Vision RAG, optional Langfuse wiring, and collection stats. Walk Through the UI The UI has four tabs: The venue-operations dashboard. It establishes the event

原文超出正文长度上限，此处截断——上游还有内容，完整版见上方「原文 ↗」。