← All content

Uptime Industries — Custom Model in Days from ~100 Datapoints

Built a custom model from ~100 datapoints in days. Self-service retraining with no vendor dependency.

Overview

Company: Uptime Industries — industrial AI for predictive maintenance and operations.

Challenge: Needed a custom NLP model for their domain but had limited labeled data (~100 examples) and no in-house ML team.

Solution: Used distil labs to train a task-specific SLM from ~100 datapoints. The platform’s synthetic data generation expanded the seed set into a full training corpus.

Results:

  • Custom model delivered in days, not weeks
  • Trained from ~100 seed examples
  • Self-service retraining with new data — no vendor dependency

“With distil labs, we built a custom model using ~100 datapoints in days. The self-service retraining has been especially valuable — we can retrain the model ourselves with new data.”

Sascha Buhrle Co-Founder & CEO at Uptime Industries