Hello everyone,
I've been closely following the discussions here about Helicone for the past few weeks as we evaluate it for our operations. We run a mid-sized manufacturing business with a fairly complex tech stack built around NetSuite for our core ERP. Like many, we initially looked at Helicone as a potential solution for managing and observing our OpenAI API usage, which we are slowly incorporating for some customer service automation and report summarization.
However, in my research, I'm seeing that nearly every review and example focuses exclusively on OpenAI's GPT models. The documentation mentions support for other providers, but I find the lack of detailed community discussion on this point quite conspicuous. My caution stems from past experiences where a tool marketed as "multi-provider" ended up being heavily optimized for one primary service, with others being an afterthought.
My question is this: Is anyone here actively using Helicone for more than just OpenAI GPT calls? I'm particularly interested in real-world use cases involving other AI providers or even completely different HTTP-based API services. For instance, does anyone use it to monitor and manage costs for Anthropic's Claude, or for vector database APIs like Pinecone or Weaviate? Perhaps even for more traditional REST APIs in a B2B e-commerce or logistics integration context?
I am trying to understand if Helicone's architecture truly allows it to function as a general observability and cost management proxy, or if that is more of a theoretical capability. Details on your configuration for non-OpenAI endpoints, the reliability of logging and metrics for those services, and any limitations you've encountered would be incredibly valuable. Our ideal scenario would involve a single pane of glass for monitoring various external API calls—both AI and non-AI—that feed into our inventory management and supply chain reporting systems, but I want to ensure the tool is robust beyond the most common use case before proposing it.