From manual consulting to the CognitX engine

Leyton occupies a specific niche in the professional services market, focusing on the intersection of government incentives and corporate research. Founded in 1997 by Jean-Pierre Giraud and François Gouilliard, the firm spent its first two decades as a traditional consultancy. It helped companies navigate the labyrinthine requirements of R&D tax credits and sustainability grants. However, as the volume of financial data grew and AI tools matured, the company shifted its strategy toward software-led automation. This transition is embodied in CognitX, Leyton’s dedicated AI and data science business unit.

CognitX is not merely an internal IT department; it is a product-focused organization within the larger firm. The engineers there, including specialists like Marouane Ben Abbou, build systems that process large-scale datasets to identify eligible R&D expenditures. This involves extracting information from payroll records, project management tools, and technical documentation to verify that a company’s activities meet strict legal definitions for tax relief. By utilizing a stack centered on Python, Django, and AWS, the team creates what are effectively financial agents capable of making expert-level determinations at scale.

The scale of global incentives

The company is headquartered in London and Paris but operates globally, with a significant technical presence in regions like Morocco. This international footprint is necessary because tax laws vary significantly by jurisdiction. Leyton’s software must be adaptable to the UK's R&D tax relief scheme, the US Section 41 credit, and various European innovation grants. This requirement for local expertise combined with centralized automation gives Leyton a competitive advantage over smaller, single-market startups. They currently employ over 1,000 professionals, a mix of tax attorneys, technical experts, and software engineers.

Competitive friction and automation trade-offs

In the competitive landscape, Leyton sits between the "Big Four" accounting firms and the newer wave of fintech startups. The Big Four offer prestige and human oversight but often at a price point and manual pace that is inaccessible to mid-sized firms. Conversely, startups like MainStreet offer highly automated, high-velocity tools but sometimes lack the deep regulatory relationships required for complex audit defenses. Leyton attempts to solve this by using its CognitX agents to handle the heavy lifting of data analysis while maintaining a global network of human consultants for high-level strategy.

This hybrid model is not without its challenges. Automating professional services requires high precision; a single error in a tax filing can result in significant penalties or audits. As a result, the development cycle within CognitX focuses heavily on validation and expert-in-the-loop systems. This ensures the AI identifies potential credits that a human might miss while relying on human experts to sign off on the final legal submissions. This tension between speed and accuracy is the defining characteristic of Leyton's current technical direction as they move toward fully autonomous financial incentive agents.