How to Intermediate · 3 min read

AI for personalized product search

Quick answer

Use LLMs combined with embedding-based semantic search and user behavior data to create personalized product search. This approach matches user queries with relevant products dynamically, improving relevance and conversion.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install openai>=1.0
pip install faiss-cpu
pip install numpy

Setup

Install required Python packages and set your OpenAI API key as an environment variable.

Use pip install openai faiss-cpu numpy to install dependencies.
Set OPENAI_API_KEY in your environment for authentication.

bash

pip install openai faiss-cpu numpy

output

Collecting openai
Collecting faiss-cpu
Collecting numpy
Successfully installed openai faiss-cpu numpy

Step by step

This example demonstrates how to build a personalized product search by embedding product descriptions and user queries, then retrieving the most relevant products using FAISS for vector similarity search and OpenAI GPT-4o for query understanding.

python

import os
import numpy as np
from openai import OpenAI
import faiss

# Initialize OpenAI client
client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

# Sample product catalog
products = [
    {"id": 1, "name": "Wireless Bluetooth Headphones", "description": "High-quality wireless headphones with noise cancellation."},
    {"id": 2, "name": "Smart Fitness Watch", "description": "Track your workouts and health metrics with this smartwatch."},
    {"id": 3, "name": "Portable Charger 10000mAh", "description": "Compact power bank for charging devices on the go."},
    {"id": 4, "name": "Ergonomic Office Chair", "description": "Comfortable chair with lumbar support for long work hours."}
]

# Embed product descriptions
product_texts = [p["description"] for p in products]
response = client.embeddings.create(model="text-embedding-3-small", input=product_texts)
product_embeddings = np.array([data.embedding for data in response.data]).astype('float32')

# Build FAISS index
dimension = product_embeddings.shape[1]
index = faiss.IndexFlatL2(dimension)
index.add(product_embeddings)

# Function to embed user query
def embed_query(query: str) -> np.ndarray:
    resp = client.embeddings.create(model="text-embedding-3-small", input=[query])
    return np.array(resp.data[0].embedding).astype('float32')

# Personalized search function
def personalized_search(user_query: str, top_k: int = 3):
    query_vec = embed_query(user_query)
    distances, indices = index.search(np.array([query_vec]), top_k)
    results = []
    for idx in indices[0]:
        product = products[idx]
        results.append(product)
    return results

# Example usage
query = "Looking for noise cancelling headphones"
results = personalized_search(query)
print("Search results for query:", query)
for r in results:
    print(f"- {r['name']}: {r['description']}")

output

Search results for query: Looking for noise cancelling headphones
- Wireless Bluetooth Headphones: High-quality wireless headphones with noise cancellation.
- Smart Fitness Watch: Track your workouts and health metrics with this smartwatch.
- Portable Charger 10000mAh: Compact power bank for charging devices on the go.

Common variations

You can enhance personalized search by:

Using async calls with the OpenAI SDK for better performance.
Streaming partial results for interactive search experiences.
Switching to other embedding models or larger LLMs like gpt-4o for richer query understanding.
Incorporating user profile data or past interactions to re-rank results.

python

import asyncio
from openai import OpenAI

async def async_embed_query(client, query: str):
    resp = await client.embeddings.acreate(model="text-embedding-3-small", input=[query])
    return resp.data[0].embedding

async def main():
    client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])
    query = "Ergonomic chair for office"
    embedding = await async_embed_query(client, query)
    print("Async embedding vector length:", len(embedding))

asyncio.run(main())

output

Async embedding vector length: 1536

Troubleshooting

If embeddings fail, verify your OPENAI_API_KEY is set correctly and has access to embedding models.
If FAISS index search returns no results, ensure embeddings are correctly computed as float32 numpy arrays.
For slow responses, consider caching embeddings or using smaller models.
Check for API rate limits and handle exceptions gracefully.

✅

Key Takeaways

Use embedding vectors and FAISS for efficient semantic product search.
Combine LLMs for query understanding with vector search for personalized results.
Leverage user context and behavior data to improve search relevance dynamically.

Verified 2026-04 · gpt-4o, text-embedding-3-small

Verify ↗