The same pattern. Every sector. Every stage.
The same pattern. Every sector. Every stage.
In fifteen years working across regulated industries in business development and enterprise sales I watched the same pattern repeat.
Post-seed AI founders making major commitments on beliefs that had never survived contact with the real world.
Not because they were careless. Not because they lacked intelligence or domain knowledge. Because they were systematically biased toward the assumptions that had carried them this far. The beliefs that got them funded felt like facts. The people who would prove those beliefs wrong were never in the room.
In fintech I watched founders build twelve months of relationship capital on champions who had no purchasing authority. In healthtech I watched clinical validation get mistaken for procurement readiness. In insurtech I watched broker enthusiasm get mistaken for underwriter acceptance. In legaltech I watched fee earner adoption get mistaken for firm-level buying intent. In defence I watched field test success get mistaken for a path to programme of record.
The product was right. The assumptions underneath the commercial motion were wrong. And by the time that became visible, the cost was already embedded in the runway.
That is the pattern. It is consistent. It is predictable. And because it is predictable it is preventable.
In fifteen years working across regulated industries in business development and enterprise sales I watched the same pattern repeat.
Post-seed AI founders making major commitments on beliefs that had never survived contact with the real world.
Not because they were careless. Not because they lacked intelligence or domain knowledge. Because they were systematically biased toward the assumptions that had carried them this far. The beliefs that got them funded felt like facts. The people who would prove those beliefs wrong were never in the room.
In fintech I watched founders build twelve months of relationship capital on champions who had no purchasing authority. In healthtech I watched clinical validation get mistaken for procurement readiness. In insurtech I watched broker enthusiasm get mistaken for underwriter acceptance. In legaltech I watched fee earner adoption get mistaken for firm-level buying intent. In defence I watched field test success get mistaken for a path to programme of record.
The product was right. The assumptions underneath the commercial motion were wrong. And by the time that became visible, the cost was already embedded in the runway.
That is the pattern. It is consistent. It is predictable. And because it is predictable it is preventable.
Why this keeps happening.
Why this keeps happening.
The reason founders carry assumption debt is not a failure of effort or intelligence. It is a structural feature of how post-seed companies move.
Two cognitive biases account for most of it.
Confirmation bias makes founders seek evidence that validates what they already believe. Every customer conversation that goes well reinforces the assumptions underneath it. The conversations that would prove those assumptions wrong never get scheduled because nobody is looking for disconfirmation when the narrative is working.
Commitment bias makes founders double down on decisions already made rather than question the assumptions underneath them. The further along a company is, the harder it becomes to see which beliefs are load-bearing and which are comfortable fictions that the team has been building on for months.
Both biases are stronger, not weaker, in founders who are intelligent, experienced, and moving fast. Domain expertise in regulated markets amplifies them further. Founders who spent a decade inside a bank or a hospital or a law firm carry deep conviction about how those markets work. That conviction is often right. And it is also the reason the assumptions underneath their commercial motion go unquestioned for longest.
The reason founders carry assumption debt is not a failure of effort or intelligence. It is a structural feature of how post-seed companies move.
Two cognitive biases account for most of it.
Confirmation bias makes founders seek evidence that validates what they already believe. Every customer conversation that goes well reinforces the assumptions underneath it. The conversations that would prove those assumptions wrong never get scheduled because nobody is looking for disconfirmation when the narrative is working.
Commitment bias makes founders double down on decisions already made rather than question the assumptions underneath them. The further along a company is, the harder it becomes to see which beliefs are load-bearing and which are comfortable fictions that the team has been building on for months.
Both biases are stronger, not weaker, in founders who are intelligent, experienced, and moving fast. Domain expertise in regulated markets amplifies them further. Founders who spent a decade inside a bank or a hospital or a law firm carry deep conviction about how those markets work. That conviction is often right. And it is also the reason the assumptions underneath their commercial motion go unquestioned for longest.
Where Assumption Crucible™ came from.
Where Assumption Crucible™ came from.
I spent time in decision science and cognitive bias research looking for a structural explanation for what I had been watching. CB Insights found 42% of AI startups fail due to insufficient market demand. MIT found 95% of generative AI pilots deliver no measurable return. Not execution failures. Assumption failures.
The research confirmed what the pattern had already shown. The problem is not that founders make assumptions. Every company runs on assumptions. The problem is that the assumptions most likely to be wrong are the ones nobody has thought to test because they feel like facts.
I built Assumption Crucible™ in May 2026 to be the thing that did not exist when those founders needed it most. A structured methodology for identifying and pressure-testing the assumptions underneath a major strategic commitment before the founder makes it. Grounded in decision science. Delivered before the damage is done.
The methodology is called the Assumption Crucible Method™. It maps assumption debt across twelve categories, ranks assumptions by how load-bearing they are and how tested they are, and runs structured external conversations with the people who hold real evidence about whether those assumptions hold.
The deliverable is a written decision memo. Not a slide deck. Not a list of things to think about. A decision.
I spent time in decision science and cognitive bias research looking for a structural explanation for what I had been watching. CB Insights found 42% of AI startups fail due to insufficient market demand. MIT found 95% of generative AI pilots deliver no measurable return. Not execution failures. Assumption failures.
The research confirmed what the pattern had already shown. The problem is not that founders make assumptions. Every company runs on assumptions. The problem is that the assumptions most likely to be wrong are the ones nobody has thought to test because they feel like facts.
I built Assumption Crucible™ in May 2026 to be the thing that did not exist when those founders needed it most. A structured methodology for identifying and pressure-testing the assumptions underneath a major strategic commitment before the founder makes it. Grounded in decision science. Delivered before the damage is done.
The methodology is called the Assumption Crucible Method™. It maps assumption debt across twelve categories, ranks assumptions by how load-bearing they are and how tested they are, and runs structured external conversations with the people who hold real evidence about whether those assumptions hold.
The deliverable is a written decision memo. Not a slide deck. Not a list of things to think about. A decision.
Regulated markets are not interchangeable.
Regulated markets are not interchangeable.
The assumption debt carrying the most risk in fintech is different from insurtech, which is different from healthtech.
In fintech the heaviest debt sits in regulatory posture, positioning against incumbents and foundation models, and inference costs that do not survive contact with production economics. FCA authorisation timelines are longer than founders model. Tier-one bank data privacy requirements are stricter than founders assume.
In insurtech the debt concentrates in champion selection, where brokers have enthusiasm and underwriters have authority, and in regulatory compliance with Lloyd’s requirements that actuarial teams need explainability the product cannot yet provide.
In healthtech the debt sits in enterprise deployment, where a three week clinical pilot meets an eleven month NHS procurement process, and in champion selection, where clinical leads with no budget authority drive the relationship.
In legaltech the debt is in pricing authority, where fee earners love the product and managing partners control the budget, and in data sovereignty, where assumptions about client data flowing through external APIs collapse in SRA compliance conversations.
In defence and govtech the debt sits in the gap between field test success and programme of record, where data sovereignty obligations and DE&S commercial framework requirements were never visible in the testing environment.
The sectors are different. The pattern is the same. Founders treating assumptions as facts until the market proves otherwise.
The assumption debt carrying the most risk in fintech is different from insurtech, which is different from healthtech.
In fintech the heaviest debt sits in regulatory posture, positioning against incumbents and foundation models, and inference costs that do not survive contact with production economics. FCA authorisation timelines are longer than founders model. Tier-one bank data privacy requirements are stricter than founders assume.
In insurtech the debt concentrates in champion selection, where brokers have enthusiasm and underwriters have authority, and in regulatory compliance with Lloyd’s requirements that actuarial teams need explainability the product cannot yet provide.
In healthtech the debt sits in enterprise deployment, where a three week clinical pilot meets an eleven month NHS procurement process, and in champion selection, where clinical leads with no budget authority drive the relationship.
In legaltech the debt is in pricing authority, where fee earners love the product and managing partners control the budget, and in data sovereignty, where assumptions about client data flowing through external APIs collapse in SRA compliance conversations.
In defence and govtech the debt sits in the gap between field test success and programme of record, where data sovereignty obligations and DE&S commercial framework requirements were never visible in the testing environment.
The sectors are different. The pattern is the same. Founders treating assumptions as facts until the market proves otherwise.
How I work.
How I work.
Assumption Crucible™ is not consultancy. Not coaching. Not advisory. It is founder-side due diligence.
Investors run due diligence on founders before they write a cheque. By then it is too late for the founder to fix what gets found. Assumption Crucible™ runs the same rigour for the founder before the damage is done. Same stress test. Different timing. Run for the founder, not against them.
I work with a small number of post-seed AI founders each year in regulated markets. Founders who are approaching a major strategic commitment and who want to know the ground is solid before they jump. The methodology requires full access to the founder’s thinking, their team, and their customers. That level of depth is not compatible with volume. Each engagement is delivered personally.
Every engagement starts with a 30 minute qualification call. If the fit is not right on both sides the call ends there and you owe nothing.
Assumption Crucible™ is not consultancy. Not coaching. Not advisory. It is founder-side due diligence.
Investors run due diligence on founders before they write a cheque. By then it is too late for the founder to fix what gets found. Assumption Crucible™ runs the same rigour for the founder before the damage is done. Same stress test. Different timing. Run for the founder, not against them.
I work with a small number of post-seed AI founders each year in regulated markets. Founders who are approaching a major strategic commitment and who want to know the ground is solid before they jump. The methodology requires full access to the founder’s thinking, their team, and their customers. That level of depth is not compatible with volume. Each engagement is delivered personally.
Every engagement starts with a 30 minute qualification call. If the fit is not right on both sides the call ends there and you owe nothing.
Why did you build this in 2026 specifically?
+
How many founders do you work with each year?
+
Do you work outside the UK?
+
What sectors do you work in?
+
Find out if you are carrying Assumption Debt™
Find out if you are carrying Assumption Debt™