AI will replace people who use it like a magic button.
It will amplify people who know how to direct it.
After a year of building AI products for sales teams in high-stakes government markets, the question I keep getting asked hasn’t changed. People want to know if AI is going to replace them.
My answer is no. It produces noise when people try to use it as a substitute for skills they haven’t yet built. The operators who learn to direct these tools come out of this transition stronger.
That conclusion comes from a year of watching AI succeed and fail in real businesses, across three completely different industries.
The Sam Altman moment that started this
Last week I watched a replay of Sam Altman comparing what it takes to build AI solutions to what it takes to raise a child to become a productive economic unit. It was painful to watch.
It was painful because it revealed the gap.
AI founders are talking about labour as an economic input. Working people are wondering whether their experience, judgment, and years of hard-earned skill still matter.
The disconnect between AI company founders and the people whose jobs are being talked about keeps widening. Someone has to answer the actual question working people are asking, and the answer should come from someone who has spent real time building these tools in real businesses.
So I want to share what I’ve learned across three different applications of AI over the past year.
How SOS Signals started as research and became a product
The first application is SOS Signals. It started as research for a book on what it takes to win in B2G markets.
I interviewed dozens of clients, partners, and people in my network who sell into government. The same problem kept surfacing across every conversation.
They knew their markets, they knew their buyers, and they knew what it took to win. What they couldn’t do was find the time to research accounts at the depth their work required.
That research was eating the hours they should have been spending on relationships and proposals. So I spent six months building structural frameworks that taught AI to deliver consistent research output across different industries and regions.
The work that mattered wasn’t the prompts. It was the operational discipline that kept the AI from producing the kind of generic output you’ve seen from every tool that promises to replace expert judgment.
The AI could process more evidence than I could manually, but it couldn’t decide what evidence mattered until I taught it how experienced operators think.
How a voice agent solved an onboarding problem at scale
The second application is a voice onboarding agent for a colleague whose business is growing faster than he can personally onboard new hires.
The challenge is one most growing companies recognize. The further the founder gets from the day-to-day, the harder it becomes to transmit the deep operational knowledge that makes new employees successful in their first 90 days.
Attrition climbs. Probationary turnover rises.
We loaded the agent with his experience, his client-specific knowledge, and the operational rules that govern his business. New hires now have access to a voice agent that responds the way he would, augmented by the knowledge of his second-in-command and his operations manager all at once.
It’s consistent in a way no individual trainer could be. It works because it amplifies the founder.
How healthcare procurement is closing a gap nobody is solving
The third application is healthcare procurement, currently in early pilot.
Sitting in a hospital waiting room with my girlfriend after she broke her toe, I used Signals to check whether anyone was working on the obvious problem of five-hour ER waits. It turns out there’s a fully funded, politically supported initiative across Canada to improve patient flow and reduce wait times.
The gap is between the demand and the small number of vendors who can deliver solutions that fit within government procurement frameworks. We’re building tools that sit on top of existing systems and help close that gap.
The tool made the pattern visible faster.
The human value was knowing where to point it.
The principle that connects all three
Three different industries, three different problems, three different agents. The principle underneath all of them is the same.
AI works as a complement and amplifier to great people. The operators who learn to direct these tools come out of this transition stronger.
Why human oversight stays non-negotiable
Even simple AI tasks fall apart without human oversight.
I asked an AI tool a basic question about my hockey team’s coach and the Stanley Cup Playoffs. It told me confidently he was in contention for an award. Two questions later, the same tool told me he was sitting out this season and wouldn’t be eligible.
The tool was lying about facts I happened to know better than it did. If you can’t trust AI to get a hockey question right, you can’t trust it to do your work for you.
What you can trust it to do is amplify what you already know how to do, when you have the experience to direct it correctly.
What this means for operators
If you’re worried AI is coming for your job, the question to ask yourself is whether you’re directing the technology or hoping it directs itself. The operators winning right now have specific, structural approaches to working with AI. They aren’t getting replaced. They’re getting more leveraged.
Don’t compete with the machine.
Don’t surrender to it.
Be the Human Boss of the technology.
Listen to the full episode of The Signal on Spotify or Apple Podcasts, or watch on YouTube:
