Finding possible integrations

[ogzhanolguncu]

There are a few LLM observability tools we can integrate, but we need to check their feasibility first. For now, we can start with:

• LangSmith
• Promptlayer
• LLMonitor
• Langfuse
• Guardrails AI
• PromptTest

Any PR that includes these tools is welcome.

[Iamsidar07]

@ogzhanolguncu Should we implement these in the example code or the source code? I want to clarify where they should be integrated.

[ogzhanolguncu]

Well, I already finished Langsmith. Most of the products above are analytic tools so we should extend our SDK with those. Right now whenever we initialize a model we do this:

const ragChat = new RAGChat({
    vector,
    model: upstash("meta-llama/Meta-Llama-3-8B-Instruct", {
      analytics: { name: "langsmith", token: "XXX" },
    }),
  });

We just need to add some more. At least check its feasibility.

[umutbozdag]

Hi @ogzhanolguncu, what do you think about integrating Cloudflare Gateway AI?

I think we can implement it within the same way that you implemented LangSmith for observability.

[ogzhanolguncu]

I agree. Feel free to give it a try.

[umutbozdag]

I have opened a PR for it @ogzhanolguncu. Can you please check? #58