I keep hearing some version of the same sentence:
“We need to use more AI.”
Usually what that means in practice is: the team bought three tools, plugged in one of them halfway, named a Slack channel after innovation, and now wonders why nothing is compounding.
This is why I keep coming back to operations.
According to the 2026 State of RevOps report from SyncGTM, 51% of sales leaders say disconnected systems are actively slowing down AI initiatives, and reps still spend just 40% of their time actually selling. On top of that, Supermetrics’ 2026 Marketing Data Report analysis says only 6% of marketers have fully embedded AI into their workflows, while 52% say data strategy and measurement decisions are made by external teams.
That is not a tooling gap.
That is an orchestration gap.
The real blocker is boring
This is the part founders hate because it sounds less exciting than “agentic workflows.”
AI only looks magical when the underlying system is clean enough to support it.
If your CRM is messy, your definitions are inconsistent, your routing is weird, your enrichment is unreliable, and your handoffs are tribal knowledge, AI does not fix that.
It accelerates the confusion.
Why RevOps matters more now
In the old model, bad operations mainly created annoyance.
A rep had to dig around. A marketer had to export a CSV. A CS leader had to ask RevOps to patch a field.
In the AI model, bad operations kill leverage.
Because now the machine needs:
clean data
clear triggers
reliable permissions
stable workflows
consistent objects and definitions
Without that, every “automation” becomes a high-speed way to make dumb decisions faster.
My founder take
I love elegant tooling.
I love a working API even more.
But I have learned that the teams who get the most out of AI are rarely the teams with the most tools.
They are the teams with the cleanest plumbing.
That is why the rise of the GTM engineer makes sense to me. The operator who can connect signals, systems, enrichment, CRM logic, and execution is becoming insanely valuable because modern GTM is increasingly a systems design problem.
What strong operators do differently
They map the motion before automating it
If nobody can explain how a lead becomes pipeline without opening six tabs and squinting, you are not ready for more AI.
They define ownership
Who owns lifecycle stages? Who owns routing? Who owns enrichment logic? Who owns lead quality? Who owns field hygiene?
If the answer is “kind of everybody,” the answer is nobody.
They reduce system sprawl
Every new tool creates another place for truth to fracture.
They build around decisions, not dashboards
The point is not “better reporting.” The point is faster, better commercial action.
A good diagnostic question
If your team says AI is underperforming, ask this:
“What exact workflow became materially better?”
Not “what tool did we buy?” Not “how many prompts did we test?” Not “how many seats are active?”
What workflow got faster, cleaner, cheaper, or more reliable?
If nobody can answer that cleanly, you do not have AI transformation.
You have AI tourism.
What I would prioritize this quarter
If I were handed a messy GTM stack, I would do this in order:
Fix lifecycle definitions
Clean routing and ownership rules
Audit data quality at the field level
Remove redundant tools
Document the highest-friction workflows
Automate only the motions with clear inputs and clear business value
That order matters.
You earn the fun stuff.
The bigger lesson
AI is exposing management debt.
For years, companies could hide sloppy GTM architecture behind hustle, spreadsheets, and heroic employees.
Now the standard is higher.
If you want AI to work like labor, your systems have to work like systems.
Bottom line
Most teams do not need more AI first.
They need better operations first.
Because once the plumbing is right, AI can finally do what everyone promised it would do: remove drag, speed up execution, and let your best people spend more time where judgment actually matters.
