John Hetherington leads the conceptual development of Cassi — the ideas and first principles beneath the product, and the questions it exists to answer. Chief Innovation Officer and a co-founder, he has spent his career at the meeting point of intelligence analysis, human judgement and the psychology of decision-making, and it is that lens he brings to what Cassi should be and where it goes next.
He came to that work through the British Army’s Intelligence Corps, serving as an Arabic and Pashto linguist and as a combat and operational intelligence analyst, with tours in Northern Ireland, the former Yugoslavia, Iraq and Afghanistan. Throughout, he held a long-standing interest in psychology and war — how people actually reason under uncertainty, and how that bears on tactical planning and execution — and presented on its application in the field.
Like Keith Dear, John had read Philip Tetlock’s work on forecasting and become a strong advocate for the approach within UK Defence. The two met after Keith’s Beheading the Hydra, and in 2015 co-authored Assessing Assessments, the paper that first made the case that would become Cassi’s direction: that intelligence assessments detached from measurable reality — a UK terror threat level implying a daily attack probability of 75% or more, held for a decade — are worse than no assessment at all.
He went on to more senior roles in Defence before reuniting with Keith and Al Brown at Fujitsu’s Centre for Cognitive and Advanced Technologies, where the team built the forecasting AI that won the Metaculus Forecasting Cup in 2024. In March 2025 they left to build Cassi full-time. John’s job there is the one it has always been: to keep the company honest about the problem it is really trying to solve.
