If I had to use one word for the advent of AI:
It wouldn’t be “exciting”.
It wouldn’t be “scary”.
It wouldn’t be “revolutionary”.
No, it would be “annoying”.
As in, I find it extremely annoying to have to learn this whole new thing and adapt my processes and systems just to stand still.
Yes, I know, not very sophisticated of me, but that’s the truth
But still, now that I have adapted (somewhat) to it, I’ll admit that some of the incremental benefits are starting to show themselves.
Not for me personally – but for my wider mission to democratise strategy, and get more founders doing it.
Put simply, AI can make strategy far easier.
And so I welcome it.
But only if its strengths and limitations are well understood – because otherwise it might make a bad situation worse.
So what are those strengths and limitations?
And how should we use it accordingly?
Here are my conclusions so far:
Limitations
The first, and most important message to take on board is this:
AI can’t tell you what your strategy should be.
I will admit that if you are a business with zero strategic focus, then getting it to tell you what to do may well be better than nothing. Sure, it’ll be “best practicey” type stuff, but many people don’t even have that figured out, so let’s not dismiss it out of hand.
But such mediocrity isn’t the goal, is it?
And mediocrity is the limit of AI-created strategy.
There are two main reasons for this.
The first is simply its lack of knowledge. This seems a weird thing to say for something with seemingly godlike levels of info at its disposal, but we must remember that it only knows stuff that has been recorded and written down – which is actually a tiny amount of reality writ large.
For example, it should be obvious that it knows far less about your business than you do – because only a tiny fraction of information about your business exists in the public domain.
Of course you can tell it a lot, but it will never be able to match your gestalt view of things – let alone your intuitive ability to prioritise information in a natural hierarchy.
(Hierarchy of information is more important than quality or quantity of information – but that’s a deep technical discussion for another day)
Point is, you should never forget that you are the authority, not it.
The second reason it can’t tell you what your strategy should be is that it’s incapable of judgement.
It doesn’t “think” – which is easy to forget given that it mimics human speech codes, leading to us giving it more human characteristics than it actually has.
(An interesting example of this can be seen with this viral clip, which demonstrates how its “false reasoning” works – making errors that no human would ever make)
Unfortunately, because of this lack of discernment, it will be able to give you a pretty convincing argument in favour of any strategy if you ask it to – which just goes to show that its enthusiasm for one thing or another is hollow and not to be trusted.
Again, you are the one whose enthusiasm is valid. (Or “more” valid anyway)
Strengths
Once we accept that an LLM is not replica of the human brain (despite appearing like one), and so has weaknesses that we don’t, we are also able to recognise that it has strengths that we don’t.
And it’s these strengths that we want to use.
Other than some very obvious things like research, which I won’t waste time discussing here, the main thing AI excels at (especially for strategy development) is this:
Pattern recognition.
And this is pretty useful, because this is exactly what strategy is: a pattern.
I’ve written thousands of words in this newsletter basically saying “good strategy has X characteristic and not Y characteristic”, and taken together these words represent the mapping of a universal pattern that strategy needs to adhere to, regardless of content.
For example if we take two classic strategy case studies like Southwest Airlines and the Nintendo Switch, we can see that although their content is completely different, their patterns are nearly identical.
And it’s AI’s ability to spot these patterns where it really becomes a weapon.
Basically it can tell you whether the thing you’re looking at is a strategy or not.
Not if it’s a good strategy.
But merely if it is one.
And this is something most humans are astoundingly bad at; constantly tricking themselves, weaselling their way out of difficult decisions, and doing everything in their power to create insipid “non-strategies” that leave them feeling good but ultimately accomplish nothing.
Pattern recognition also helps us in strategy development.
You see the process we go through to arrive at a strategic breakthrough is also a pattern. At least the one I’ve developed is. And it’s a pattern that AI can carry us through with focus and discipline – where we would be at risk of veering off at a tangent.
We might summarise the division of responsibilities like this:
- The human is responsible for the content of the strategy.
- The AI is responsible for the structure of the strategy
This actually means that we must use AI in precisely the opposite way most of us tend to.
Instead of :
Human prompts.
AI answers.
It should be:
AI prompts.
Human answers.
The prompt provides the structure – the frame of the conversation. The answer provides the content. The meat. The thing that needs to be judged with a sensitive human eye.
So here’s what I’ve done
To test these theories, I’ve built the first and only strategy bot to work exactly this way:
- It prompts you
- You answer
- It shapes
- You judge
…all based on a colossal amount of training I’ve given it off the back of my no-bs process (so it’s fast, fun, and contrarian – not just useful).
As one of my early testers said:
“OK, I’m dying, it even SOUNDS like you!”
(Not quite sure how good a thing that is, but at least it’s not generic)
I believe this back and forth makes the most of the unique abilities of AI, and the unique abilities of you as the founder.
In early testing the results have been frightening.
People are able to arrive at a solid hypothesis, laid out in full strategy document form, in as little as 20 mins.
Yes, they need some extra shaping from me and the founder. The creative leaps aren’t of the quality and originality the human can do. But that’s no big deal, because the creative leaps aren’t the issue – it’s the structuring where people struggle.
And now, they don’t have to.
So far the AI coach has only been available to members of the full Strategy Shortcut System community.
But I’ve now introduced a mid-tier program – Shortcut Guided – to let others play with it at a cheaper price. It also includes one 15-minute recorded teardown of your strategy from me. So you can produce, get snap feedback, adapt, and arrive at what is by far the most cost effective strategy development program on the planet.
