How to Implement AI Search on a Webshop?

It seems like overcomplicating things! Searching within your own site is something that's been figured out for ages, and given that you have full data access, you might not need to go the AI route.

Consider using a vector database like FAISS along with an embedding model like BERT. You can pick it up in a week and set it up easily with Python and Langchain. Just keep in mind that the underlying idea that words and phrases are vectors in a meaning space has its limitations, so optimizing beyond that can be tricky.

Many retailers are moving towards AI-powered search, especially those using a hybrid approach that mixes semantic search with traditional filtering. This can significantly improve how products are discovered, especially for ambiguous searches. For instance, Walmart uses a hybrid embedding system, and it’s been really effective in driving engagement and conversion rates. Conversational chatbots are also gaining traction for boosting sales and enhancing customer interaction!

I'm looking into this too, working with a major e-commerce platform. To be honest, while AI can be cool, it's costly and doesn't always lead to increased sales. A good scenario might be using AI search when traditional search results don't match user queries well, but even then, the expenses might outweigh benefits. It could enhance user experience but won't necessarily bring in more sales.

Why not just ask ChatGPT? It can give you some ideas based on what it understands you’re looking for!

You might want to combine embeddings from OpenAI with a vector database like Pinecone. This could help link user queries and product descriptions for more context-aware search results. It’s a great way to enhance user experience with smarter results!