Tractable: The AI Startup That Taught Computers to Assess Vehicle Damage

Back to: Top 10 British Startups 2026

Every year, insurers around the world process hundreds of millions of claims for vehicle damage — the crumpled bonnets, shattered windscreens, and scraped panels that are the inevitable consequence of putting billions of cars on the world's roads. For most of the industry's history, assessing this damage has required trained human assessors to physically inspect vehicles, a process that is slow, expensive, and inconsistent. Tractable, a London company founded in 2014, has built artificial intelligence that can do the same job from photographs — faster, cheaper, and with measurable improvements in consistency and accuracy.

Company Overview

Tractable was founded by Alex Dalyac and Razvan Ranca, who met as mathematics students at École Polytechnique and went on to study machine learning at UCL and Oxford respectively. Their founding insight was that the problem of assessing damage from images was formally similar to problems that computer vision had already made enormous progress on, and that the insurance industry's tolerance for slow, manual processes represented an opportunity for technological disruption of substantial commercial value.

The company spent its first three years in pure research mode, building and refining the computer vision models needed to identify damage types, severity levels, and repair costs from photographic evidence. This investment in technical foundations has paid dividends: Tractable's models now operate with accuracy levels that have been validated against human expert assessments in independent testing, giving insurers the confidence to use AI-generated assessments as the basis for claims decisions without routine human review.

Business Model

Tractable sells its damage assessment platform to insurance companies, vehicle manufacturers, and automotive dealers on a per-assessment pricing model. The value proposition is straightforward: a Tractable assessment costs a fraction of a human assessment, delivers results in seconds rather than days, and produces consistent outputs that reduce disputes between insurers, repairers, and customers. For large insurers processing millions of claims annually, the cost savings run to tens of millions of pounds per year, making even a relatively premium pricing model highly attractive.

The company also generates revenue from its deployment in vehicle remarketing — helping dealers and auction houses assess the condition of used vehicles at scale — and from partnerships with vehicle manufacturers who use Tractable's technology to support roadside assistance programmes and warranty claim processing. These adjacent applications of the core damage assessment capability represent significant incremental revenue opportunities that the company is actively developing.

Innovation Factor

The core innovation at Tractable is a computer vision system trained on an enormous dataset of vehicle damage images, each annotated with expert assessments of damage type, severity, and repair cost. This training dataset, accumulated over years of commercial deployment and continuously expanded through new partnerships, constitutes a significant competitive moat — it would take a new entrant years to assemble comparable training data, and without it, the accuracy levels needed to satisfy insurer requirements are essentially unachievable.

Beyond the base damage assessment capability, Tractable has developed specialised models for specific damage types — glass claims, total loss assessment, flood damage — and is expanding into structural damage assessment for property insurance. Each new domain requires substantial domain-specific training data and validation work, but the company's existing relationships with major insurers provide both the data and the validation infrastructure needed to move into new verticals efficiently.

Market Position

With a $1 billion valuation and deployments at more than 20 global insurers including Ageas and Aviva, Tractable is the established market leader in AI-powered vehicle damage assessment. Its technology is now embedded in the claims workflows of insurers representing hundreds of millions of policies worldwide, giving it a level of operational integration that makes displacement by a competitor extremely difficult. See also: Healx's AI in healthcare and Faculty AI's enterprise AI deployments.

What's Next

Tractable is expanding its technology beyond vehicle damage into property and casualty insurance more broadly, applying its computer vision expertise to assess storm damage, flood damage, and fire damage from aerial and satellite imagery as well as ground-level photographs. This expansion opens an addressable market considerably larger than automotive alone, and the company's track record of technical accuracy and operational reliability gives it a credible foundation for this next chapter of growth. Visit tractable.ai for more information.