Deep learning with
real world impact.

Discovery process illustration

Breakthrough discoveries to real products

We carry out meaningful applied research: transforming breakthrough discoveries in deep supervised and semi-supervised learning into products that help people recover from accidents and disasters more quickly.

Vehicle analysis illustration

Assessing damage is a visual task

With the right approach and training, this can be carried out by an AI like a human, at scale.

Automation illustration

Addressing complex legacy processes

To get there we are addressing complex processes that, today, experts with decades of experience take hours to complete.

Damaged truck illustration

Understanding real world damage

Our research focuses on creating representations that capture the subtlety of real damaged vehicles and properties.

Noisy data illustration

Building robust training methods to deal with data that is noisy, heterogeneous and of varying dimensionality.

Refining illustration

Refining domain adaptation techniques to handle differences in geographies and repair standards.

Combining illustration

Combining visual output with fine-grained metadata to utilise key non-visual information.

Conditioning illustration

Conditioning learning on structured expert knowledge to create models that are accountable and interpretable.

Want to help?

We are building AI-first products that will transform traditional industries by streamlining outdated processes and delivering unprecedented customer value.

Join us

Get updates