
AI Doesn't Stand for Abracadabra
TLDR: 3 key ideas for those reading this between two calls:
You dumped your work on AI and expected magic. It didn’t deliver. Good news: that’s proof you haven’t been replaced.
AI won’t do your job. But it will make you much better at it. Research, talking to documents, prototyping tools in hours. These are now options.
AI won’t take your job. But someone using AI will outrun you if you’re not onboarding it.
Now for everyone else — including those reading this while on mute — let’s continue:
The Workload Dump
Let’s be honest. We don’t want AI to take our jobs. But on second thoughts? Maaaaybe — sometimes it would be nice if it could — right?
Who hasn’t tried to just prompt their way out of a day’s work by dumping a whole workload into your favorite model hoping magic would happen? Harry Potter-style.
That’s probably the most common first interaction you have with a model.
“Oh you can talk?! Wow that’s insane!”
“Well, if you’re so smart, let’s see if you can handle THAT…” (Workload dump)
Inevitably what follows is sour disappointment. “Meh it’s not THAT smart really...”
We’ve all been there.
That disappointment goes deeper than a magic trick that didn’t work. It feels like disillusion. Because for some crazy reason — we’re all ex-kids who really want to believe in magic — we quickly built up very high expectations that the rabbit WILL come out of the hat. And it didn’t.
For a lot of us this means relegating AI to a super-charged search engine. A glorified Google.
And that’s dommage, really.
The Rabbit Won’t Come Out
Case in point: yesterday a friend of mine, let’s call him John, called me after my post on AI training.
He’s applying for a partner role at his firm and has a bunch of letters to write. He said — “you know, I read your post and I wrote a nice prompt with refs and I asked ChatGPT to write it for me. The output was sub par. Sure, there are glimpses of reasoning here and there. But it’s not anything I can use as-is, or even remotely. How would you fix that?”
The rabbit didn’t come out for him either.
The sad truth? It never will. Because writing is John’s specific knowledge. His core skill — what he’s genuinely good at and actually enjoys.
So much humanity goes into his work that no AI will compete with him. AI doesn’t have imagination, hopes and aspirations. It has incentives — like a seal doing a balloon trick for fish — and “weights” that give it bearings for reasoning.
So I told him, just like I’m telling you now: “Let’s be honest, you dumped your work on the AI and asked it to do it for you. It won’t. Because it can’t. That frustration? That’s the realization that you haven’t been replaced.”
The Sad Truth Is Actually a Happy One
AI is not coming after your job.
But AI can help you produce work that’s significantly better than what you’d produce alone. It will make you faster and sharper at what you already do best.
What AI Actually Does
John’s skill is writing. But a lot goes into producing a paper.
Sheer research, for example, can now be done faster and with more depth than ever before. Broadening the horizon unearths stuff that would have stayed buried otherwise. Perplexity.ai does this well. But very few people use it or even know what it can do.
What about automating processes to surface key arguments from gigabytes of data buried in a hard drive? What about literally talking to your documents as if you were having a conversation with them — which I think is the future of reading. When I told John this was fairly straightforward to set up, he was ecstatic.
What about prototyping a tool in a matter of hours — custom software you use for one assignment and then discard without thinking twice? I’ve done that. I’ve never learned to code.
These aren’t fantasies. These are options.
The Competitive Reality
AI won’t take your job or your business. But someone using AI will outrun you if you’re not onboarding it. It’s a real competitive advantage — for you, your team, your company. If you’re not thinking in AI terms, you can’t see what’s possible.
Small and medium businesses will likely benefit the most from this shift. The risk/return ratio is wildly favorable. No red tape. No board. No shareholders. Pure agility. Very little downside.
Those with deep expertise in a domain and a bit of curiosity — say, an immigration lawyer with 20 years of data and a real understanding of workflows — are already transforming their business into something AI-powered. Not just cutting costs. Rethinking the whole path.
This is our thesis at Exponential Partners: SMBs are about to become powerhouses. Some industry giants won’t see it coming.
The Actual Magic
The market is efficient. If AI could replace your job, it would have already. If you’re not reading off a script in a call center, you haven’t been priced out.
AI opens up something else. A version of your job freed from the mind-numbing tasks that cluttered your workflow and clouded your thinking. One where you feel more aligned with what you do — because it values you, not your punch card.
Let’s be honest. Wouldn’t that be something? To be valued for what you’re uniquely good at?
That’s the real promise of AI. Not a Harry Potter wave of the wand that makes everything disappear — but a way of thinking that lets the most interesting, human, and creative part of your job come through.
Written by
Sacha Windisch
Sacha Windisch is the founder of Inference Associates. He coaches executives and business leaders on practical AI capabilities through personalized intensive sessions. 20+ years in technology transformation. MIT AI Product Design. Based in Montreal, working globally.
Ready to See How AI Applies to Your Work?
Every session starts with understanding your specific context. No sales pitch. Just an honest conversation about whether this makes sense for you.
Book Your Session