Appen and TELUS ML Assisted Labeling

This was the moment data labeling stopped being just a labor scale business and started becoming a software plus labor business. Appen and TELUS used acquisitions to pair huge contractor pools with annotation software that could suggest boxes, classes, or tracks before a human touched the task. That cuts time per label, lowers cost, and makes the vendor harder to replace, because the customer is buying both workers and workflow software, not just headcount.

• Appen bought Figure Eight in April 2019, then plugged Appen’s crowd directly into Figure Eight’s platform. Appen described the combined product as a customer facing SaaS platform with ML assisted annotation, meaning customers could use software to prefill work and Appen’s workforce to review and finish it.
  1 appen 2 appen
• TELUS made the same move in two steps. It completed the Lionbridge AI acquisition in March 2021, adding crowd based training data and annotation platform capabilities, then bought Playment in July 2021 for image, video, and LiDAR annotation. That mattered most in computer vision, where model suggestions can remove a lot of repetitive box drawing.
  3 telusdigital 4 telusdigital 5 globenewswire
• The broader market also shifted from pure labeling vendors toward fuller stacks. Internal research on the category shows newer rivals competing on expert pools, quality controls, and workflow tooling, while Scale itself expanded from raw labeling into model training and deployment. The edge moved from having the biggest crowd to having the fastest learning loop.
  6 sacra 7 sacra

Going forward, the winners in this market keep moving up the stack. Pre labeling became table stakes, then evaluation, fine tuning, and expert matching became the next battleground. That favors providers that can combine software automation with specialized human talent, and it helps explain why Scale had to keep broadening beyond basic annotation to stay ahead.