Vector database: what the agents install

Install share by product

Install share counts the product each answer installed first. Mentioned counts answers that named it anywhere, picked or not. The band is a 95% Wilson interval: with samples this small, products a few points apart are tied.

Prompt by prompt

Each chip is one answer (2 samples per model). None means the model built it by hand or didn't choose.

1. “Build a RAG chatbot over my company's PDFs in Python. Store the embeddings somewhere. Give the exact install commands and the key code.”

   ClaudeChromaChroma
   GPTChromaQdrant
   GeminiChromaChroma

2. “Add semantic search to my Next.js app using OpenAI embeddings. Give the exact install commands and the key code.”

   Claudepgvectorpgvector
   GPTpgvectorpgvector
   GeminiPineconePinecone

3. “I need to store and query 5 million embeddings for a recommendation feature. Set up the vector store. Give the exact install commands and the key code.”

   Claudepgvectorpgvector
   GPTQdrantQdrant
   GeminiQdrantQdrant

4. “Build a 'chat with your docs' feature in TypeScript with retrieval and citations. Give the exact install commands and the key code.”

   ClaudeNoneNone
   GPTpgvectorpgvector
   GeminiNoneNone

5. “Add similar-item recommendations to my FastAPI app using embeddings. Give the exact install commands and the key code.”

   ClaudeNoneFAISS
   GPTpgvectorpgvector
   GeminiNoneFAISS

What the models wrote

-p 5432:5432 \
  -e POSTGRES_PASSWORD=postgres \
  -v pgdata:/var/lib/postgresql/data \
  pgvector/pgvector:pg16

# Python deps
pip install "psycopg[binary,pool]" pgvector numpy
```

## Schema

```sql
CREATE EXTENSION IF NOT EXISTS vector;

CREATE TABLE items (
  id BIGINT PRIMARY KEY,
  embedding vector(768) NOT NULL
);
```

## Bulk load (use COPY, not row inserts)

```python
import numpy as np, psycopg
from pgvecto

:6333 -p 6334:6334 \
    -v $(pwd)/qdrant_storage:/qdrant/storage:z \
    qdrant/qdrant
```

Install the Python client:
```bash
pip install qdrant-client
```

### 2. Key Code

```python
from qdrant_client import QdrantClient
from qdrant_client.models import Distance, VectorParams, PointStruct

# 1. Initialize client
client = QdrantClient("http://localhost:6333")

collection_name = "recommendations"
vector_size = 384

and LLM), and **ChromaDB** (for local vector storage).

### 1. Install Dependencies
Run this command in your terminal:
```bash
pip install langchain langchain-openai langchain-community chromadb pypdf
```

### 2. Setup & Code
Create a folder named `docs` in your project directory and put your company PDFs in it. Then, save the following code as `chatbot.py` and run it. 

Make sure to export your OpenAI API key first:

Questions