Last week a founder pushed back on a build price.
"That feels like a lot for a setup I could probably figure out myself," they said.
I didn't argue. I asked one question instead.
"What's your time worth, and what happens to the work after you stop paying attention to it?"
Silence.
Then: "I mean... I'd have to learn it first."
And there's the whole thing. The price was never the question. The question was always what they could actually point a tool at and finish.
So last Sunday I ran the experiment on myself.
What I actually did
I spent about three and a half hours rebuilding the technical foundation of my own site. Not a redesign. The plumbing.
Technical SEO: collapsed the page to a single clean H1, added a canonical tag, added structured data so search engines know who and what this site is, fixed the social share image so links stop looking broken when someone posts them.
AEO: set the robots policy so AI answer engines can read the site to cite it, while training bots stay out. That's a real distinction now, and most sites get it wrong by accident.
CRO: I killed a dead-end "Learn more" link, reworded a tagline that wasn't doing its job, and replaced a multi-step booking flow with one click that opens my calendar. Fewer steps, fewer drop-offs.
Then I deployed a corrected sitemap, set up Search Console, verified the domain, submitted the sitemap, and requested indexing.
Three tools did the labor. Claude Code made the edits. Perplexity's Comet assistant drove the screen to handle the Cloudflare and Search Console work. Search Console did the indexing. Combined subscription cost: about $40 a month, and this used a sliver of it.
What that work costs if you hire it out
I priced it the way I'd price it for a client. Conservative. Nothing inflated.
The work spans a technical SEO audit, implementation, AEO configuration, CRO fixes, and analytics setup. Hired out in the 2026 US market, a technical audit alone runs anywhere from a few hundred to a few thousand dollars. Bundle the audit with implementation, AEO, CRO, and Search Console setup, and a freelancer or small agency lands it somewhere between $1,500 and $5,000 as a one-time project.
I took the floor of that range on purpose. Fair value of the deliverable: $1,500.
So the headline writes itself. About $1,500 of professional work. About $40 in tools. Three and a half hours.
There’s a name for what that ratio measures: Return on AI Labor, or ROAILTM, a term coined by Alicia Lyttle. It asks a simple question. For every dollar you spend on AI tools, how much real work comes back out?
On this task, the answer was about 37 to 1. And that's before you count everything else those same tools do for me all month.
That's a great number. It is also the part everyone gets wrong.
Here's what the number hides
That $1,500 is money I didn't spend. It is not money I earned.
On a real scorecard it goes under "costs reduced," not "revenue." If I let myself blur those two, I'm doing exactly the thing this newsletter exists to push back on: dressing up a real result as a bigger one.
So I'll say it plainly. A professional might have done some of this faster than I did, or caught something I missed. The honest version of the story keeps that in.
But the more interesting thing the number hides is the input nobody put a price on.
The tools didn't decide to collapse the H1. They didn't know my booking flow was leaking people. They didn't choose to let answer engines in and keep training bots out. They didn't scope the job, sequence it, or know what "done" looked like.
I did. Every one of those was a judgment call I made before a tool touched anything.
That's the uncosted input. The $40 bought execution. The judgment was already mine, and it's the only reason $40 of execution turned into $1,500 of finished work.
The sequence underneath it
This wasn't me throwing tools at a screen. It followed the same path I run with every client.
Assess: I found the actual problems, not the ones I assumed. The site had a structure issue, a booking-friction issue, and an AI-policy issue. Three real problems, named.
Optimize: I fixed the process before adding anything shiny. Removing booking steps and clarifying a dead link isn't a tool. It's a better decision about how the page should work.
Automate: only then did I let the tools execute, including the one-click calendar popup and the agentic browser handling the Search Console setup.
Run that in the wrong order and you get a fast, confident mess. You can automate a broken booking flow in thirty seconds. It just routes people off your site faster.
Why this matters for you
If you've been told AI will do your work for you, here's the correction from someone who just lived it.
AI did the labor. I did the deciding.
You can't delegate what you can't define. I could hand $40 of tools a clear, scoped, sequenced job because I already knew, in specific terms, what the job was. Most people skip that part, point a tool at a vague intention, and then wonder why the output needs a full rewrite.
Here's the part the ROAILTM number really proves. The return didn't come from the subscription. It came from the judgment I pointed at it. Same $40 of tools in different hands, with no clear definition of the job, returns a fraction of that, or a mess. The leverage isn't the tool. It's the judgment you bring to it.
One thing to try this week
Pick one task you keep meaning to "use AI for."
Before you open a single tool, write down what done looks like. The actual criteria. What has to be true for you to call it finished and walk away.
If you can write that down, the tools will earn their keep.
If you can't, that blank page is your real project. Not the tool.
Simple.
Not easy.
Worth it.
Jay Founder, Clarity2Scale Consulting
Process-First AI Strategist
The Clarity Team
A note of thanks: the Return on AI Labor (ROAILTM) framing comes from Alicia Lyttle, "The Queen of AI" and CEO of AI InnoVision. It gave me a clean way to measure what I already believed. Credit where it's due.