Case Study: Appen achieves 97% automated fraud detection with Provectus

Appen boosts fraud monitoring 20x with Provectus

The customer, Appen, a global AI training data company, faced the challenge of monitoring its crowdsourcing platform for fraudulent contributors. With a manual system that could only check 50 annotation jobs per day, they were unable to scale and needed an automated solution to prevent low-quality data from poisoning client datasets. They partnered with vendor Provectus to build an ML-powered fraud detection platform.

Provectus implemented a machine learning system on AWS that automated the detection of malicious behavioral patterns. The solution monitored over 1,000 jobs per day with 97% automation, leading to a 25% reduction in scammer activity and a 5x decrease in wasted annotations. This allowed Appen to avoid hiring 20+ data analysts and ensured higher quality data for its enterprise clients.

Appen