01
What AI is really worth
Why licences, usage and time saved reveal activity while saying far less about changed organisational capability.
AI expert · keynote speaker · independent researcher
Richard Foster-Fletcher
Companies cannot yet see what their AI can really do.
Richard studies frontier models, enterprise AI systems and the decisions being made around them. He brings that work to major conferences, boards and senior executive rooms.
LinkedIn Top Voice in AI, 2024 · Favikon UK Top 20 AI Researcher, 2026 · Author of The AI Gap, Kogan Page, June 2027.
The work
The people accountable for AI may never see what it can really do.
The models available outside a company can differ sharply from the systems permitted within it.
Richard examines what changes between frontier capability and enterprise deployment, what those changes mean for consequential decisions, and what becomes difficult to see once systems, policies and commitments form around a restricted version of AI.
Speaking
What Richard speaks about
The talks begin with the technology itself: what frontier systems can now do, where they remain unreliable, and what changes when their capability is narrowed for enterprise use.
02
How AI can strengthen judgement
Where advanced models can extend analysis, test competing explanations and alter the understanding behind a consequential decision.
03
How AI changes what an organisation knows
What happens when models begin shaping the information, summaries and reasoning on which organisational decisions depend.
Independent research
Where the story and the evidence part company
Richard studies companies from the outside, comparing public claims with policies, contracts, system behaviour and reported results.
The work often begins with something widely accepted that does not quite make sense, then follows the evidence across organisations until the underlying pattern becomes visible.
Visit the MKAI research archive →404 : 2Public statements about AI adoption and containment, compared with statements about changed capability, across 22 large companies.
Adoption is easier to report than capability.
Companies can count use while remaining unable to establish whether AI changed what the organisation could understand or decide.
Governance changes the capability being governed.
The controls required for deployment can remove the directness, range and instability from which much of the model's capability arises.
Agents extend the existing path.
They increase the speed and reach of deployed systems while carrying many of the same underlying capability limits.
_1783702420594-C5awYkhH.jpg)
The book
The AI Gap
Organisations were offered AI as intelligence and deployed it as software.
The book examines the distance between the capability companies believe they have acquired, the capability their systems permit, and the responsibility senior figures continue to carry.
The wider work
Research, writing and named mechanisms
↗
Research
Companies report AI adoption far more readily than they report changed organisational capability.
↗Writing
The most confident AI claims often change once implementation begins producing its own evidence.
↗Structural Dynamics
AI changes organisations through recurring mechanisms that conventional software language cannot explain.
↗Ask the Analysis
The published archive answers enquiries within the limits of its documented evidence.
Speaking enquiries
What AI can do, what enterprise systems change, and who carries the result.
Richard speaks at major conferences, board sessions and private senior briefings on artificial intelligence and the decisions that shape a business.
Book Richard to speak