Case Study: Bos-Dimex achieves offline text classification with Inoxoft

Bos-Dimex builds offline text classification with Inoxoft in 10–15 minutes

Inoxoft developed the WhiteLightning open-source AI tool to address a challenge for Bos-Dimex. The customer needed fast, cost-effective, and offline text classification for applications like spam filtering and content moderation, avoiding the high costs and privacy risks of cloud-based LLMs.

The solution was a command-line tool that distills a large language model into a compact, sub-1MB ONNX classifier. Inoxoft's implementation resulted in a model that is 95% cheaper to run, processes over 2,500 texts per second, and operates entirely offline on devices from servers to Raspberry Pi hardware.

Bos-Dimex