2028: As a (fictional) CEO, I came under pressure two years ago to produce a strategic plan by a sharp deadline. While I accept it’s part of my job, all of a sudden I received not-too-subtle demands from the board, funders, staff, and even regulators. They all expected to see a written document.
The problem was, no-one on my executive team wanted to set the time aside to meet, let alone make the financial investment. Then I read an article from Harvard Business Review and it hit me. I could use AI.
While the authors of “Researchers Asked LLMs for Strategic Advice. They Got ‘Trendslop’ in Return” highlighted the shortfalls of using AI to write a strategic plan, I saw the upside.
Within a few minutes, I could feed an AI/LLM a bunch of data related to my company’s performance, and the wider industry. I could get all the PESTER stuff I wanted and even get the LLM to produce a decent SWOT Analysis, using a range of inputs.
The end-result?
All those people who expected to see our strategic plan by the due date were satisfied. Not only did I far exceed the deadline, but they didn’t even ask followup questions. Of course, they could have been too busy to absorb it properly.
But I think that even those who clicked it open and took a glance had no issues. While our team didn’t intend to consciously implement much of it, the thinking inside the document was logically sound. And beautifully formatted. And well-researched. Professional.
Want to know why this shortcut “worked?” When we hired a real strategy consultant two years later and finally did the right thing, I realized why I got away with such a drastic shortcut. And, why I won’t make that move again.
1) A 5-Year (i.e. Short-term) Focus Made It Easy. Simple.
In retrospect, I can own the fact that the document AI wrote wasn’t really a strategic plan. In spite of its name, and the liberal use of the right buzzwords, I now see it differently.
The truth is that the document didn’t amount to more than an extension of our C-Suite’s Monday-morning huddles.
Perhaps this was no accident, because I fed it all the relevant presentations and meeting notes from the past two years. That biased it towards our most recent discussions.
But here’s the kicker. I asked the LLM to write a five-year strategic plan.
That’s the norm for most companies, but it meant that the plan it produced was no more than a prioritized list of problems we were already aware of. The AI merely extended these priorities into the near future, imagining that we would continue to try to solve them.
Here and there, the AI introduced wildcards such as a nascent technology and a new industry regulation we were preparing for. But these were glancing blows. Neither would be disruptive in the five-year horizon I asked the AI to focus on.
Notwithstanding this limit, the AI did a better job at examining the totality of current issues than any of our executives could do on their own. Its perfect memory and lightning execution were admirable.
But, as I came to learn, this approach didn’t make the plan strategic. Not all plans are as I just learned from Roger Martin’s popular YouTube video on the subject. https://youtu.be/iuYlGRnC7J8?si=Rfj1dl8kHeDyT7hv
First Lesson – My initial takeaway is that the AI plan was simple because it only focused on solving existing problems. As such, it didn’t live up to the challenge our consultant proclaimed: “If it ain’t game-changing, then it ain’t a strategy.”
To all who read it, there was little to argue with. There was also nothing to be inspired by. Now we have a 20-year vision and a strategy to go with it.
2) Focusing on the Document Led Us to Focus on Logic Only
As I mentioned, I originally thought the LLM did a good job. But back then, I thought the purpose of a corporate strategy process was to produce a written plan.
Now, I see the mistake. A project to craft a real strategy always has a social dimension. A social outcome.
Yes, there’s also an intellectual dimension. This is what I was taught in my strategy course in MBA school.
But in the real world, it’s folly to craft a strategy that doesn’t include stakeholders of all kinds. It’s also a mistake to overlook the fact that the process is not like a relay, where each person does their part along the way, leaving a single person to complete the final leg.
Instead, it’s more like competitive rowing where everyone in the boat crosses the finishing line together.
From this experience, I learned that corporate strategy is a team activity. But it’s not a sport which involves physical effort.
Instead, it’s a social and intellectual project involving a core group in which participants actively engage each other. As Robert Twemlow puts it:
“Stronger judgment is forged through encounter: with problems, with constraints, with evidence, with other people, and with the practical realities that force thinking to become more precise.”
While an AI or consultant can write a credible, professional document, this by-product is secondary. Apparently, few of my stakeholders knew the importance of a new level of team cohesion when I sent out the strategic plan for comment.
The Second Lesson is that strategy is far more than a document. It’s a new compact between C-Suiters and their stakeholders in all directions. If there’s no fresh challenge being imagined with a collective, intellectual rigor, then it’s not likely to include a strategy within its pages.
3) The Hybrid Trap — It Let Us Avoid Every Hard Trade-Off
The AI’s plan was full of “and” instead of “or.” We’d differentiate and cut costs. Centralize core functions and empower the regions. Pursue radical innovation and incremental improvement.
The HBR researchers found this same pattern: when LLMs are allowed to dodge a binary choice, they recommend “do both” 24% of the time. It reads as sophisticated. Michael Porter would say it leaves a firm stuck in the middle.
Looking back, the plan went unchallenged precisely because it asked nothing of anyone. No budget was redirected. No initiative was killed. No executive had to defend a sacrifice in front of their peers.
Third Lesson: Strategy is sacrifice. If your plan doesn’t say what you’re going to stop doing — what customer you’re walking away from, what product you’re killing, what region you’re exiting — it isn’t a strategy. It’s a wish list.
And this failure to contend with tradeoffs means that no-one owns anything.
When our consultant arrived a few weeks ago and asked: “Who owns the differentiation push? The regional restructuring? The partnership strategy?” the room went quiet. Not because we were dodging blame, but because no one had ever made the choice. Thanks to an LLM, the plan “emerged”. It was no one’s wager.
Perhaps this was the reason it was “accepted” so widely, without even a whisper. There just wasn’t any accountability.
4) What the Shortcut Actually Cost
Two years on, I can name what that shortcut actually cost. None of it appeared on any P&L.
We never had the argument about whether to walk away from our second-largest customer segment. We had it twenty-four months later, with the consultant in the room, and it was the most useful three hours my executive team had ever spent together. The AI plan had quietly recommended we serve those customers better. Nobody had been forced to choose.
We never built the muscle of staking a position and defending it. My CFO and my COO have a real disagreement about whether we should grow by acquisition or by building internal capability. It has been simmering for three years. The AI plan papered over it with a hybrid recommendation that pleased everyone in the moment and committed nobody to anything afterwards. The disagreement is still simmering. We are still paying the interest on it.
We never earned the trust of the board. They accepted the document, but I now realise they didn’t believe in it — they simply had nothing to push back against. Trust between a board and a CEO is built by watching the CEO defend a hard call. I gave them nothing to watch.
I got my bonus. I made my deadline. I kept my chairman happy. And I postponed every conversation that strategy is supposed to force into the open.
That is the real ledger of an AI-written strategy: not the costs you can see, but the conversations you didn’t have, the bets you didn’t place, and the muscle the organisation didn’t build. By the time you notice them missing, the year is gone.
5) My Learning: The Real Failure Was Mine, Not the AI’s
The hardest thing for me to admit, writing this more than a year later, is that the AI didn’t fail. It did exactly what I asked of it. I requested a document; it produced a document. I asked for a five-year plan that would satisfy the board; the board was satisfied. I asked it to summarise where we stood and project forward; it summarised and projected.
The failure was upstream of the machine. By the time I typed that prompt, our company had already turned strategy into a document-production exercise. The AI didn’t deskill us — it revealed how deskilled we had quietly become. The shortcut was already drawn into the floor of this company by years of treating the binder as the deliverable. The LLM just walked the path we had worn.
That is the part I want every person reading this open letter to sit with. AI will expose, ruthlessly, the strategy practice you actually have — not the one in your charter or your annual report. If your practice is mostly about producing a document on time, an LLM will produce one for you, and the lights will stay on, and the bonuses will land, and nothing real will change. Until something does, and you find you have nothing in the room to meet it with.
Part of the reason I’m writing this open letter (which my stakeholders are sure to read) is to apologize.
That is the door I am closing on this company — not on AI, but on the kind of strategy practice an AI can complete on our behalf. Such tricks should never enter our company again.
From here on, if a conversation can be replaced by a prompt, then it wasn’t a strategic conversation in the first place. Let’s be vigilant to make sure this never happens again.