I’ve been working on a project that involves calling LLM APIs from different providers—Watsonx.ai, AWS Bedrock, and OpenAI.
Pretty quickly, I ran into a challenge:
Each provider has its own SDK, its own API interface, and its own way of doing things. So even for a simple task like making a model call, I had to dig into separate docs, learn different SDKs, and write separate chunks of code for each.
It was slow. And it didn’t feel like the best use of time.
Then I came across AI Suite—an open-source Python library from Andrew Ng. It acts as a thin wrapper over the official SDKs from all the major providers, but with a unified interface. That means I can switch from one provider to another just by changing one line of code.
It made the whole workflow a lot smoother.
If you’re curious, I’ve also shared my notebook with working examples here:
👉 GitHub: https://github.com/shaikhq/genai-labs/blob/main/aisuite-sample/
