It's fascinating that when we think intensely about something, we suddenly start noticing it everywhere. A good friend who studies psychology tells me it's a known phenomenon (the Baader-Meinhof effect).
Another known phenomenon is that (too!) much parenting happens in the car. During one such trip, listening to my daughter's playlist, a line from Billie Eilish's song particularly stuck with me: "Don't waste the time I don't have." Not because it was exceptionally insightful, but because it perfectly illustrated my feelings and the atmosphere in company boardrooms.
Regardless of whether I'm talking to the board of a large corporation, a private equity fund's portfolio company, or the investor themselves, the question is usually similar: how can we use AI without wasting time we already have too little of?
The answer I've arrived at after two decades on the business side is uncomfortably simple: most AI programs fail not because of the technology. They fail because they were a pilot, not a system. Solutions that continue to work years later rarely stem from a single brilliant idea. And in the case of AI: from a single use case.
What truly remains after several years is a system of interconnected actions, embedded in the organization's way of functioning. Such a system has its own checks and balances. The board receives clear information about progress and results. There are clear metrics for success. Employees understand both the goal and their role in achieving it. Over time, if well-designed, it begins to operate almost autonomously.
Therefore, the most important question about AI, one that is thoroughly operational and outcome-focused, is not: what use case should be implemented, but how to build something that will stand the test of time.
Why most AI implementations don't lead to lasting change
These are first-hand observations. If we look at organizations before the start of an AI program, we very often see the same picture:
- AI is on the board's agenda, but few leadership teams have had a joint, structured conversation about what AI truly means for their organization and what ambitions they want to achieve.
- Ideas are plentiful. Employees have their own initiatives, vendors present new solutions, and managers watch new technology demonstrations. The problem is that the value of these ideas is rarely assessed in a consistent and comparable manner. This 'unclarity' does not help boards make quick and accurate decisions.
- AI very often ends up in technology departments. Tools appear in the organization, but the operating model remains unchanged. Over time, old habits return, and assumptions about return on investment begin to blur.
- Organizations collect use cases instead of selecting them. Ideas multiply, but there is a lack of filters to assess business value, data readiness, implementation feasibility, or adoption potential. As a result, it is difficult to demonstrate impact on company results.
- Many organizations begin their journey without analyzing the current state. Without an honest assessment of the current level of maturity, it is difficult to decide where we want to be in a year and what level of ambition is realistic.
- The phenomenon of Shadow AI develops. When employees do not receive approved tools and clear guidelines, they resort to their own solutions. Some experiment without adequate safeguards, while others wait for decisions and remain passive.
- Skill development remains the domain of a small group of enthusiasts. A few people gain experience very quickly (often outside the organization) while the rest fail to keep up with the change.
I suspect that most readers will find their organization in at least some of these observations. Crucially, none of these problems stem from the technology itself. These are challenges related to leadership, decision-making, and the organization's mode of operation. This is also one of the reasons why so many companies, despite significant investments, still do not see a lasting impact of AI on business results.
Implementing AI is an operating model challenge, not a technology one
In discussions about AI, we often hear that one should start with strategy. This sounds almost like a revealed truth, although in practice, strategy can be an abstract document written "for the drawer" or prepared for an audit.
Personally, I prefer to start with something simpler: an honest, data-driven conversation about where the organization is today. Only after looking at the facts can one consciously decide on the level of ambition. What position do we want to occupy? How fast do we want to grow? What does the regulatory environment look like? What are investors' expectations? In which areas do we want to compete and how do we intend to build an advantage?
For such conversations, at Symmetria Partners, we use our own AI maturity model. In addition to data quality and technology, it includes leadership, employee engagement, organizational readiness for change, reporting methods, and AI's place on the board's agenda.
+44% AI maturity level in four months. This was achieved in one regulated organization, and we could demonstrate it so precisely precisely because we started by measuring the baseline.
Interestingly, most organizations that began their AI journey several months ago are still at a relatively low level of maturity. Calling things by their name is usually the first step to making the right decisions.
How to build value from AI that remains after the project ends
Around a clearly defined level of ambition, a system of actions is built that allows the organization to develop AI even after the transformational program concludes.
The first element is an organized process for evaluating use cases.
40+ initiatives analyzed → 3 implemented. In one program, three initiatives were selected for implementation that had the potential to translate into tangible business value within a reasonable timeframe.
Added to this are common standards for tool usage, a secure environment for experimentation, a community of engaged employees, and communication tailored to the organizational culture. Reporting is no less important: boards, supervisory boards, and investors expect transparent information on progress, maturity level, competence development, risk, and achieved results.
Ultimately, however, the most important thing is transferring responsibility to the organization. After many years of operational work, I am convinced that this is the moment that distinguishes a project from a true organizational capability.
A system of interconnected actions instead of individual use cases
None of these elements are particularly unique on their own. Value appears only when they start to drive each other. The inherent interdependence between connected actions is the true implementation power.
It's easiest to see this as a cycle:
- Thorough use case evaluation selects initiatives with real business value.
- These initiatives feed reporting with concrete, measurable results.
- Results keep AI on the board's agenda, so the topic doesn't disappear after initial enthusiasm.
- Presence on the agenda protects the budget and mandate for further development.
- The budget funds employee competencies and community.
- Competent employees generate further valuable initiatives, and the cycle begins anew.
Such systems are viable. Great pilots without reporting disappear from the agenda, and competencies without a continuous influx of good use cases wither. A single pilot can generate impressive opportunities, but not a steady stream of ideas that generate successive 'wins' year after year. That's why systems win. Additionally, from a leadership perspective, such a system conveys vision of actions, consistency, clear direction, organizational support, and reduces organizational resistance (to AI), which is ubiquitous in companies.
This is especially important in portfolio companies. Ultimately, what matters is not the number of experiments conducted, but the organization's ability to systematically create value within the investment horizon and the organizational skill to ensure repeatable successes.
Returning to Billie Eilish's words: "Don't waste the time I don't have," the reason why it's worth building mutually supportive systems rather than focusing solely on individual use cases is quite simple. Only then can one be sure that in the rush of AI implementation, the organization won't waste the time it already has so little of.
In my experience, this is what differentiates programs that are hard to remember a few years later from those that permanently change how a company operates and are reflected in its value. To the benefit of everyone: the board, investors, and employees.
And how does it look in your organization?
Is AI in your organization still a collection of pilots, or already a system? If you want to start with an honest assessment of your starting point, write to us.
At Symmetria Partners, we begin every AI transformation with a maturity assessment.
