How to set up a local development AI stack with LangChain, Postgres, and Weaviate

What this does

This guide provisions a local development environment for building AI agents using LangChain, PostgreSQL (with pgvector) for structured data and vector storage, and Weaviate as a dedicated vector database for semantic search. Each component runs in Docker Compose with persistent volumes, and the Python environment connects to all services. This stack mirrors a production AI architecture at development scale.

Steps

1. Create the project directory and Python virtual environment:

   mkdir ai-dev-stack && cd ai-dev-stack
   python3 -m venv .venv && source .venv/bin/activate
   pip install langchain langchain-core langchain-community pgvector psycopg2-binary weaviate-client
   

2. Create docker-compose.yml with Postgres (pgvector) and Weaviate:

   version: "3.8"
   services:
     postgres:
       image: pgvector/pgvector:pg16
       environment:
         POSTGRES_USER: aiagent
         POSTGRES_PASSWORD: secret123
         POSTGRES_DB: ai_agents
       ports: ["5432:5432"]
       volumes: ["pgdata:/var/lib/postgresql/data"]
       healthcheck:
         test: ["CMD-SHELL", "pg_isready -U aiagent"]
         interval: 5s
     weaviate:
       image: semitechnologies/weaviate:1.25.0
       ports: ["8080:8080"]
       environment:
         AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: "true"
         PERSISTENCE_DATA_PATH: "/var/lib/weaviate"
         ENABLE_MODULES: "text2vec-transformers"
       volumes: ["weaviate_data:/var/lib/weaviate"]
   volumes:
     pgdata:
     weaviate_data:
   

3. Start the services and verify they are healthy:

   docker compose up -d && docker compose ps
   

   Expected output: both services with status healthy.

4. Initialize the pgvector extension in Postgres:

   docker compose exec postgres psql -U aiagent -d ai_agents -c "CREATE EXTENSION IF NOT EXISTS vector;"
   

   Expected output: CREATE EXTENSION.

5. Configure LangChain to connect to both backends. Create config.py:

   PG_CONNECTION = "postgresql+psycopg2://aiagent:secret123@localhost:5432/ai_agents"
   WEAVIATE_URL = "http://localhost:8080"
   

6. Initialize the Weaviate client and create a schema for document embeddings:

   import weaviate
   client = weaviate.connect_to_local(host="localhost", port=8080)
   client.collections.create(
       name="Documents",
       properties=[{"name": "content", "dataType": ["text"]}],
       vectorizer_config={"vectorizer": "text2vec-transformers"}
   )
   

7. Initialize the LangChain Postgres vector store for conversation history:

   from langchain_community.vectorstores import PGVector
   from langchain_community.embeddings import OllamaEmbeddings
   
   embeddings = OllamaEmbeddings(model="nomic-embed-text")
   vectorstore = PGVector(
       connection_string=PG_CONNECTION,
       embedding_function=embeddings,
       collection_name="conversation_history",
   )
   

8. Test the full pipeline: store a document in Weaviate, index conversation history in Postgres, and run a LangChain chain that queries both:

   doc_coll = client.collections.get("Documents")
   doc_coll.data.insert({"content": "Sample document for testing"})
   vectorstore.add_texts(["Conversation turn 1: User asked about deployment."])
   
   from langchain.chains import RetrievalQA
   weaviate_retriever = doc_coll.as_retriever()
   pg_retriever = vectorstore.as_retriever()
   # Combine retrievers in the chain
   

Verification

curl -s http://localhost:8080/v1/.well-known/ready | jq

Expected output: {} (Weaviate ready, HTTP 200).

Common failures

• pgvector extension fails to create — ensure the Docker image is pgvector/pgvector:pg16, not the standard postgres:16 image which lacks the vector extension.
• Weaviate schema creation hangs — the text2vec-transformers module downloads a model on first use. Wait for the download (visible in docker compose logs weaviate) before creating the collection.
• LangChain cannot connect to Postgres — the PG_CONNECTION string must use psycopg2 (not psycopg2-binary) in the driver portion. Confirm connectivity with psql -h localhost -U aiagent -d ai_agents.

Related guides

• Create a Docker Compose stack with vLLM, Redis, and Prometheus
• Implement healthchecks and restart policies for AI containers
• Set up a multi-service AI logging stack with Docker Compose