Encord Multimodal AI Data Platform

Encord, operated by Cord Technologies Limited in London and Cord Technologies, Inc. in the United States, builds a data platform for teams training computer vision and physical AI models. It tackles the cost and slowness of managing, curating and labeling large volumes of images, video and sensor data as models move from prototype to production.

The platform covers annotation and labeling for images, video, audio, documents, DICOM and NIfTI files and 3D LiDAR point clouds, with AI-assisted labeling using Segment Anything Model 2, object tracking, consensus review and annotator analytics. Index handles data curation with embedding-based search, outlier and duplicate detection, while Active supports label error detection and model evaluation. Data stays in customers' AWS, Azure or Google Cloud storage. More than 300 AI teams use it, including UiPath, Woven by Toyota, Skydio and Synthesia.

• Multimodal annotation: labels images, video, audio, documents, DICOM and 3D LiDAR in one workflow.
• AI-assisted labeling: Segment Anything Model 2, model prediction import and object tracking.
• Index curation: embedding search, outlier detection and duplicate detection across up to billions of files.
• Active evaluation: label error detection, model analytics and model comparison.
• Quality workflows: consensus review, annotator training and performance dashboards.
• Data-as-a-service: managed annotation and data collection for physical AI.

Vietnamese data annotation service providers can run large image, video and LiDAR projects for overseas clients on a single quality-controlled platform. Automotive and robotics teams can curate edge cases from fleet data. Hospitals and medical AI startups can label DICOM imaging under HIPAA and GDPR-aligned controls.

Raw data stays in the customer's own cloud storage, avoiding data migration. AI-assisted labeling and model-in-the-loop curation reduce manual annotation and surface rare edge cases. UiPath reports 10x dataset growth and a 4x reduction in error rate using Encord. Enterprise options include SSO, US or EU hosting, private VPC and on-premise deployment.

Computer vision and physical AI projects in robotics, autonomous vehicles, drones and smart cities depend on large multimodal datasets, and data quality has become the main bottleneck in moving models to production. Vietnam hosts a sizeable data annotation workforce serving global AI teams, and local AI developers increasingly need tooling that combines labeling, curation and evaluation.

Operated by Cord Technologies Limited (London) and Cord Technologies, Inc., with teams in London and San Francisco. Raised a USD 60 million Series C in February 2026 led by Wellington Management, with Y Combinator, CRV, N47 and Crane Venture Partners; total funding USD 110 million. Completed a SOC 2 examination and states HIPAA and GDPR compliance. Data on the platform grew from 1 to over 5 petabytes in 12 months.