This repo uses Go to embed text, upsert it to a vectorDB and then query it.

Specifically:

• The input is Whatsapp chat history
• The embedding model is OpenAI's ada-002
• The vector DB is Pinecone

1. Obtain an OpenAI Api Key
2. Obtain a Pinecone API Key
3. Save a Whatsapp chat history at the path "./en_files/en_chat.txt"`` The expected format is [09.09.23, 14:35:02] ~ john_doe: Hello world!`
4. Run go run main.go
5. Follow the instructions - choose action embed/upsert/query and then a language, current options are he/en. Adding languages simply means another prefix ot the input file name in the case block at main.go

• No tests here, which is not a recommended practice.
• Currently the message text is not stored in the vectorDB. This means that when querying - you get the nearest vector, but not the associated text, which is the search result. This is marked as a TODO. What needs to be done is change the code so that the upsert includes the original text in the metadata for each entry.
• Neither OpenAI nor Pinecone have an official Go client, so it's all cURL commands. Here's where the debug-commands.txt comes in handy.
• I am doing this because I think Go is a great choice for AI applications. Benchmarks can be great to prove this point, but are not part of this repo.

• If you want to make a contribution or reuse the code: go ahead!
• If you want to contact me about this, I'm mostly active on Twitter