Case Study: Geberit achieves 5x faster data labeling with HumanSignal's Label Studio

A HumanSignal Case Study

Preview of the Geberit Case Study

Geberit boosts email labeling to 93% accuracy with HumanSignal

Geberit, a major European supplier of sanitary technology and bathroom ceramics, faced significant challenges with manual data annotation for its AI models. The process was cumbersome, time-consuming, and inefficient, particularly for intricate datasets like emails. Seeking to leverage generative AI, Geberit attempted an in-house integration with GPT-3.5 but found it prone to inaccuracies and high costs. The company turned to its existing vendor, HumanSignal, for help using their Label Studio Enterprise platform to automate and improve the process.

HumanSignal implemented a multi-faceted solution using Label Studio Enterprise, integrating advanced AI models and introducing a "Labeling Copilot" feature for semi-automation. This included guardrails to ensure output quality and eliminate AI hallucinations. The results were substantial: Geberit achieved 93% accuracy relative to human-reviewed labels, a 5x faster throughput, and 4-5x lower cost savings across LLM charges and annotator time. HumanSignal's solution empowered Geberit's subject matter experts and transformed their data processing efficiency.


View this case study…

HumanSignal

17 Case Studies