Good models are not enough. They should also be easy to build with.
The model is only the first interface
A model can perform well and still be hard to use. Developers also have to handle authentication, request formats, files, retries, errors, and the differences between one endpoint and the next.
We spent the past few months improving that layer. The result is the Addis AI SDK, available as an official package for Node.js and Python.
Node.js
npm install addisai
Python
pip install addisai
One package across the stack
The SDK covers text generation, text-to-speech, speech-to-text, and translation. It uses the same Addis AI API key across those capabilities, so a team can move from a playground test to code without rebuilding the integration each time.
The point is not to hide what the API does. It is to make the common path obvious and keep the details consistent in both languages.
1
Create an API key in the developer platform.
2
Install the Node.js or Python package.
3
Choose chat, voice, transcription, or translation.
4
Ship the same integration into a real product.
Developer experience is infrastructure
More than 800 developers made over 15 million Addis AI API calls in ten months. That usage made the next constraint clear. Access to a model is useful, but a predictable way to build with it is what lets a developer keep moving.
The SDK is our answer to that part of the work. Better clients, clearer documentation, and fewer integration decisions make the models more useful without changing the models themselves.
Start with the package
The quickstart covers API keys, installation, and first requests in Node.js, Python, and cURL. The SDK packages and docs will continue to move with the APIs as we add capabilities.