Everyone is asking the same question right now:
If AI can do everything, what happens to service businesses?
If you run an agency, consulting firm, outsourcing company, or dev shop, you've probably already heard the narrative. AI will automate knowledge work. Customers won't need vendors. Everyone will have a team of agents doing the work for them.
I don't think that's what happens.
But I also don't think AI is just hype.
Anyone who has spent real time with these tools knows they can produce useful work in minutes that used to take days. But as a friend said, its impressiveness is inversely proportional to your competence.
So if AI isn't replacing service businesses or even jobs, what is it actually changing?
What do you actually sell?
It's worth separating two very different kinds of service businesses.
Many service firms are selling labor at a margin.
Staff augmentation, contract staffing, and body-shopping businesses are fundamentally selling access to people. The customer manages the work, defines the priorities, and owns the outcome. The vendor is primarily supplying capacity.
That model has existed for a long time because hiring is slow, specialized talent is scarce, and organizations often need flexibility.
But AI puts real pressure on this category.
If one engineer can produce the output that previously required several people, customers need fewer external resources. If AI reduces the amount of routine work required to deliver a project, buying additional labor becomes less attractive.
By contrast, outcome-oriented service businesses are in a different position.
Customers hire them because they want systems that work, products that ship, revenue that grows, and problems that get solved. The customer isn't buying hours. They're buying accountability.
I think AI accelerates the decline of pure staff augmentation while increasing the value of firms that can reliably own and deliver outcomes.
What AI Actually Changes
What AI changes is the speed at which outcomes can be delivered. Execution is getting dramatically faster.
Things that used to take days now take hours. And when execution speeds up, something interesting happens. A different bottleneck emerges.
The New Bottleneck: Coordination
Most organizations spend more time waiting than working: waiting for requirements, feedback, approvals, handoffs, or someone else to finish before they can start. Historically, slow execution masked these delays, but AI is changing that. As execution time collapses, coordination becomes the primary bottleneck, making organizational friction impossible to ignore.
Why Productivity Gains Require a New Coordination Model
That's why I think the biggest misconception is that productivity gains will come from individuals becoming more productive.
If AI makes individual contributors dramatically faster but the coordination model stays the same, much of that gain gets trapped inside existing workflows.
People still spend enormous amounts of time waiting for specifications, approvals, feedback, and handoffs.
The queue is still there.
In fact, as execution gets faster, coordination becomes even more obviously the bottleneck.
Without changing how work moves between people, faster execution doesn't necessarily translate into faster outcomes. It just means more processing time between sprints.
That's why the real opportunity isn't just AI as a productivity tool for individuals, but developing with more interesting and probably proprietary coordination models.
The Rise of Multiplayer AI
What gets interesting is when multiple people can operate inside the same AI-mediated workflow.
Imagine a product manager, an engineer, a designer, and a customer all interacting with an agent/harness in a shared context.
The moment work is completed, the next person can engage. And maybe before work begins, they can better agree on what they need. This is how things have been built; we can now just compress some aspects of that work.
No formal handoffs, status-update delays, or information loss between teams. The AI isn't replacing the people involved. It's reducing the friction between them. In many cases, it's replacing the queue.
Instead of accelerating a single worker, it accelerates the movement of work through the entire system.
What Customers Actually Care About
And that's important because customers ultimately care about cycle time, not whether you used AI. They care that features ship faster, problems get solved faster, decisions get made faster, and outcomes arrive sooner. The value isn't in the technology itself but in reducing the time between identifying a problem and delivering a solution.
The Service Firms That Will Win
The service firms that win in the AI era won't be the ones with the most AI tools—they'll be the ones that redesign their operating model around shorter cycle times. AI speeds up execution, multiplayer AI speeds up coordination, and once both bottlenecks shrink, the real insight becomes obvious: customers aren't buying labor, they're buying how quickly you can turn a problem into a solution.