AI is making speed the standard (and slow teams will pay)
Photo by Jake Weirick on Unsplash
I’ve avoided talking about AI too much.
Mostly because I don’t want to be another person shouting “AI!” at the internet.
But I’ve been thinking about it a lot recently, and I keep landing on the same point:
Speed is becoming the standard.
If you’re slower, you won’t just feel it “one day”.
You’ll feel it everywhere.
Where slow shows up first
- Sales. Prospects get used to fast turnaround. Your timelines start to sound like excuses, even if your work is good.
- Delivery. More waiting. More hand-offs. More “we’ll get to it next week”. More stress that didn’t need to exist.
- Margin. Busywork is expensive when someone else has found a way to cut it down.
- Hiring. Good people don’t stick around in teams where everything takes ages and nothing ships cleanly.
That’s why I think founders need to take this seriously. Not because AI is cool. Because the pace of work is changing.
The first wins are boring (and that’s good)
The most annoying part is the first wins are boring. It’s admin. It’s the repeating tasks. It’s the stuff you keep meaning to fix and never quite get around to.
Used well, the speed jump is real. 2–5x isn’t unusual on the right jobs.
So here’s what I’d do if I was a founder trying to be sensible about it.
The simplest way to start (without doing anything reckless)
Pick one workflow with the most moving parts.
Not the easiest task. The messy one.
The one that crosses tools, involves judgement, breaks in edge cases, and leaves a trail of Slack messages and half-finished spreadsheets behind it.
Depending on your business, that might be:
- onboarding a new client from “signed” to “work started”
- weekly reporting that pulls from three places and still ends in a debate
- invoicing that needs timesheets and attachments done “just so”
- cleaning messy data before it hits finance
- anything supplier-related where email chains multiply and nobody knows what’s actually agreed
Then treat the model like a new hire.
Don’t give it the keys to the building. Start small. Review the output. Correct it. Watch where it breaks. Increase access over time.
Same way you’d bring a junior into your IT team. You don’t hand them admin access on day one. You build trust and boundaries.
If you want a couple of concrete examples of what this looks like in practice, I wrote them up here:
https://www.jonmunson.co.uk/posts/changing-how-i-work-in-ops-2026-with-ai/
Try one workflow this week.
If you want a hand choosing the right one and setting it up safely, drop me a message.
