More than 180 AI evaluation specialists delivered 3.49 million annotations across image, video, audio, and stylized-content workflows, supported by workflow-specific calibration and a 24-hour client feedback loop.
Chicago, Illinois--(Newsfile Corp. - September 24, 2026) - Hugo Inc. has completed a large-scale annotation and evaluation program spanning eight XR workflows, delivering approximately 3.49 million annotations at 93.4% accuracy.
Hugo delivered 3.49 million annotations across eight XR workflows, including motion, audio, image, and anatomy evaluation.
To view an enhanced version of this graphic, please visit:
https://images.newsfilecorp.com/files/11664/315839_2a19bc46843b2b03_001full.jpg
The program supported AI systems built for realistic XR avatars, in which motion, facial expressions, audio clarity, eye tracking, segmentation, and stylized human anatomy each pose a distinct evaluation challenge.
Segmentation required spatial AI specialists to apply up to 40 labels to stylized images with blurred edges, blended colors, and distorted anatomy. Free-form motion captioning introduced a new challenge: two trajectory evaluation specialists could describe the same movement differently, yet both be valid, creating apparent inconsistencies during QA review.
More than 180 AI evaluation specialists worked across the program, supported by workflow-specific guidelines, training, and daily quality reviews. Tasks ranged from trajectory and motion evaluation and video tagging to skeletal-mesh evaluation, acoustic evaluation, segmentation, and OCT whole-eye evaluation. A 24-hour root cause analysis loop translated client feedback into next-day guideline updates, allowing teams to resolve ambiguity before it became a recurring quality issue.
"Two specialists can describe the same motion differently, and both be right. The mistake is assuming that divergence is a rater problem when the rubric hasn't clearly defined what consistency should look like," said Sean Neighbors, Strategic Accounts Executive at Hugo Inc.
"You cannot solve that by telling 180-plus specialists to simply 'be more consistent.' You need a feedback loop that identifies ambiguity quickly, resolves it with the client, and turns disagreement into clearer guidance before it becomes a recurring pattern."
The program delivered:
- 3.49 million annotations across eight workflows
- 93.4% accuracy against an 85% target
- 85.3% utilization against an 85% target
- 1,289 seconds average handling time against a 1,974-second target
- 34.7% less handling time than the target
The work sits within Hugo's Omni Evals practice, which covers human evaluation across perception, embodied AI and robotics, speech and audio, and multimodal grounding.
Hugo will discuss the operational systems behind high-reliability AI evaluation at its Bay Area event with Frontier Labs on September 29 in San Francisco.
The event will bring together engineering leaders, GenAI researchers, and AI leads during Bay Area AI conference week for a discussion on AI operations, human evaluation, and the infrastructure required to support increasingly capable AI systems. Registration is open at https://luma.com/hugo-bay-area.
About Hugo Inc.
Hugo Inc. is a next-generation customer experience and operations partner delivering services across Customer Experience, Back Office Support, Trust & Safety, and AI Operations. Powered by a global workforce of more than 4,500 people across five countries, Hugo helps leading organizations scale high-quality operations through a combination of exceptional talent, operational expertise, and innovative technology. Hugo was named one of the fastest-growing customer-service BPO companies worldwide for the third consecutive year in 2026.
Media Contact:
Name: Andie K., PR Director, Hugo Inc.
Email Address: andie@hugotech.co
Company Website: hugoinc.com/
To view the source version of this press release, please visit https://www.newsfilecorp.com/release/315839
