A founder starts reading things more closely than they used to. A client proposal, a board paper, a market summary from someone in the team. Nothing is clearly wrong with any of them. But a sentence sounds right and can’t be traced to anything, or a figure is close to the real one without being it. So the next document gets the same careful read, and the one after that.
Nobody decided this. It builds up gradually and sits in the founder’s week next to everything else, usually without a name.
It can be the first visible sign that a team is using AI unevenly. There is no incident to point to and no project that failed. What changes is how much of the team’s work needs checking before it can be trusted, and who ends up doing the checking.
The obvious reading
The usual response is to treat it as a skills gap. Some people use the tools well, some barely use them, and a few use them with more confidence than care. So the company runs a workshop, shares a few prompts, and expects the spread to narrow.
That helps, but it treats the issue as individual, and the cost doesn’t stay with the individual. Work passes through several hands before it leaves the building: someone drafts, a manager edits, a head of function approves, the founder signs off. If one person in that chain accepts AI output without testing it, everyone after them inherits the error, usually without knowing where it came from.
The people downstream compensate. A careful manager rewrites a section; a senior colleague double-checks a number. That work is real, yet it appears in no plan or budget, and it falls hardest on the strongest people. The reliability of the team’s output ends up set by its least careful step, which is what the title means.
Why it lands on leadership
Three things tend to sit underneath this, and an individual getting better at prompting solves none of them.
There is no shared standard. Most teams adopted AI one person at a time, each finding their own way. Ask what “good enough to send” means when AI has touched the work, and the answers in one room will differ. Without a common answer, everyone applies their own.
Review depends on understanding. A manager who doesn’t use these tools well has trouble judging what was made with them. They usually do one of two things: approve too much, or check everything. Both cost the company something, and neither shows up until later.
The founder is often somewhere else. Founders tend to try new tools early and with real enthusiasm, while their leadership team may be ahead in some areas and behind in others. Decisions made at different levels of understanding pull in different directions, and the founder is left to reconcile them. It is the pattern described in Why Scale-Ups Slow Down: Leadership Becomes the Bottleneck, arriving by a different route.
Questions worth asking
These are not an audit. They are three questions a leadership team can work through in one conversation.
- Who decides how AI is used in each function, and do they understand it well enough to judge the output?
- When AI has helped produce a piece of work, what happens to it before it reaches a customer or the board? Who checks it, and against what?
- Is the leadership team itself at roughly the same level? If not, which decisions are being made by people who can’t fully evaluate the tools involved?
If the answers differ from one executive to the next, that difference is the finding.
The hiring question
When a gap like this becomes clear, the first instinct is often to hire for it: a Head of AI, a new COO, a senior person who will “own” the topic. Sometimes that is right. More often it is early.
If the real issue is that nobody has been given ownership of the standard, a new executive arrives into the same ambiguity. The role may be unclear, the authority may overlap with existing leaders, and the founder remains the person everyone checks with. Settling the mandate first decides whether a hire is needed at all, and what kind of person the role requires. It is how we approach any leadership decision: validate the decision before looking for the person.
If you do hire, assess for judgement rather than familiarity with tools. Tools change quickly. What lasts is knowing when output can be trusted, when it can’t, and how to build that habit in other people.
A more useful way to look at it
A team’s AI capability is usually described by its best users. Day to day, it is set by the distance between them and everyone else, and by whether anyone is responsible for that distance. That makes it a question about ownership and standards, which sit with leadership, more than a question about software.
If you recognise the founder who has quietly started checking everything, the useful first step is usually small: decide who owns the standard, and say so. Whether anyone needs to be hired comes after that. It is the kind of decision we help founders work through before any search begins.
Related Insights
The first manager layer is a scale-up’s real operating test
A strong specialist is promoted to manager because they know the work, deliver consistently and have earned the team's trust. Six months later, they are still solving the hardest problems themselves, their team waits for direction, and the founder is being pulled into disagreements that should have been handled much closer to the work. The promotion made sense. What was missing was a clear definition of the new role.
Read More
When “good performance” means something different in every team
Growth makes inconsistency visible. A product lead may reward speed, a commercial lead may reward persistence, and an operations lead may put reliability first. Each standard can make sense on its own. The trouble starts when people move between teams, managers make promotion decisions, or the board asks whether the organisation is ready for its next stage of growth. The company needs a clearer understanding of what good performance actually means—and whether that understanding is shared across the leadership team.
Read More
What progression and pay decisions reveal about a scale-up
The first difficult pay conversation often arrives before a company thinks it needs a formal framework. A strong employee has grown beyond their original role, another has been hired at a higher market rate, and a manager promises a promotion to keep someone engaged. Each decision seems manageable on its own. Taken together, however, these decisions begin to define how the company values contribution, recognises capability and decides what it is prepared to pay for.
Read More
Before hiring for a skill gap, find where the work breaks
A strong specialist is promoted to manager because they know the work, deliver consistently and have earned the team's trust. Six months later, they are still solving the hardest problems themselves, their team waits for direction, and the founder is being pulled into disagreements that should have been handled much closer to the work. The promotion made sense. What was missing was a clear definition of the new role.
Read More
Leadership commitments hidden inside a Series A plan
A Series A plan can look straightforward on paper: revenue targets, a product roadmap, hiring plans and market expansion. But behind those numbers sit assumptions about leadership. Every growth target requires greater decision capacity, every new function requires meaningful ownership, and every expansion plan assumes the organisation can carry more complexity without sending everything back to the founder.
Read More
When European expansion exposes leadership weaknesses
A company can perform exceptionally well in its home market and still discover that it is not ready to operate across several European countries. Expansion often exposes weaknesses that were hidden by proximity, informal decision-making and founder involvement. The question is no longer simply whether the company can enter another country. It is whether the leadership model can travel with it.
Read More
Why an experienced corporate executive can fail in a scale-up
The CV looks right. The executive has led large teams, managed significant budgets and worked across several markets. Yet six months later, both sides are frustrated. Sometimes the hire was wrong. But often, the problem is not the executive's experience. It is the environment in which that experience was built — and whether that experience translates to a scale-up.
Read More
Why Scale-Ups Slow Down: Leadership Becomes the Bottleneck
As scale-ups grow, hiring feels like the logical next step. Yet in many cases, the issue isn’t who you add, but how decisions still move. Leadership becomes the bottleneck long before it’s recognised.
Read More
Leadership Problems Rarely Show Up as Failure
Growth can mask structural strain In founder-led companies, leadership problems rarely present themselves clearly. Revenue is growing. Customers are satisfied. The numbers look broadly fine. From the outside, there is no visible reason to worry. That is exactly why the issue is often missed. When nothing breaks, strain is explained away as normal growth. Slower […]
Read More