Everyone's talking about AI replacing services. Sequoia says the next trillion-dollar company will be a software company pretending to be a services firm. HBR says professional services are getting eaten alive. They're both right.
And they're both missing the point.
Sequoia and HBR are thinking about this from the outside in. From the investor's chair. From the disruption playbook. But if you've actually run a services business for 25 years, you see something different.
The agencies that survive this aren't the ones bolting AI onto broken processes. They're the ones who finally understand what they were always supposed to be.
Let me explain.
INTELLIGENCE VS. JUDGMENT
Sequoia breaks work into two buckets. Intelligence work follows rules. It's process, execution, and pattern matching. Judgment work requires taste, experience, the ability to read a room or a situation, and the ability to make a call no playbook covers.
AI is now good enough to handle the intelligence side. Not tomorrow. Now.
That means every agency sitting on a pile of process-heavy, repeatable work just lost its moat. If your value proposition is "we have people who do the thing," you're competing with software that does the thing faster, cheaper, and without PTO requests.
But here's what the think pieces won't tell you: most agencies don't actually know which bucket their work falls into.
They think it's all judgment. It's not. Eighty percent of what most agencies deliver is intelligence work dressed up in nice decks and status meetings.
The judgment part? That's maybe 20% of the engagement. And it's the only part that matters.
THE CRAFT & CARE PLAY
This is where I land differently from Sequoia and HBR.
They see services as a market to disrupt. I see services as a craft to elevate.
When you strip away the intelligence work and let AI handle the execution layer, what's left is the thing agencies were supposed to be doing all along: caring about the outcome more than the output.
Not caring in the soft, feel-good way. Caring in the "I won't let you ship that because it's wrong for your business" way. Caring in the "we're going to have the hard conversation about your positioning because your revenue depends on it" way.
That's not automatable. That's not a copilot feature. That's a human who's been in the room enough times to know what the room needs.
THE REAL SPLIT ISN'T AI VS. HUMANS
It's agencies that own their judgment vs. agencies that are still selling hours of intelligence work.
If you're a founder-led agency and your business model depends on bodies doing process work, you have maybe 18 months before your clients figure out they can buy that from a platform for a fraction of what they're paying you.
But if you're the agency that says "we own the 20% that actually moves the needle, and we use every tool available to make that 20% sharper," you're not getting disrupted. You're becoming more valuable.
Because here's the thing nobody in Silicon Valley wants to admit: AI doesn't have taste. It doesn't have conviction. It doesn't wake up at 2am thinking about your client's go-to-market because something felt off in the last call.
People do that. Specific people. The ones who give a damn.
WHAT THIS MEANS IF YOU RUN AN AGENCY
Stop selling time. You're competing with machines on that now.
Start selling outcomes that require judgment. Price on the value of being right, not the cost of being busy.
Use AI to handle everything that doesn't require you to be in the room. Automate the intelligence layer ruthlessly. Not to cut costs. To make space for the work that actually matters.
And most importantly: get honest about which parts of your service are intelligence and which are judgment. Because if you can't tell the difference, your clients will figure it out for you. And they won't be gentle about it.
The agencies that win the next decade aren't the biggest. They're not the cheapest. They're the ones where the founders still care enough to do the hard work that no model can replicate.
That's always been the play. AI just made it obvious.