“Let’s try one more Wix AI roulette.”
That is an actual sentence I said.
Importantly, I said it after we already had quite a lot of evidence suggesting that trying Wix AI again was probably not the correct strategic decision.
But optimism is apparently renewable.
Executive Compass, though, did not start with Wix.
It started with a business plan.
You know the document.
The one you write to demonstrate that you have thought everything through, understand exactly what you are doing, and have a clear plan.
The business plan had other plans.
What started as a fairly straightforward executive finance proposition began asking much bigger questions.
How do executives actually make decisions when the answer is not obvious?
What happens when the numbers are technically correct, but they are telling an incomplete story?
How do you separate evidence from assumption?
And how do you know when a conclusion that looked perfectly sensible three weeks ago should probably now be thrown away?
That became Project ATLAS.
The principle behind ATLAS was fairly simple.
Make the claim clear.
Find the evidence.
Understand where the evidence stops being reliable.
Try to prove yourself wrong.
Decide how confident you actually are.
And, if the facts do not cooperate, change the rule.
Construct.
Evidence.
Boundary.
Falsify.
Calibrate.
Revise.
At the time, I thought I was designing a decision framework.
I didn’t realise I was about to use it repeatedly on my own business.
Then I needed a website.
This should have been the easy part.
I already had a premium Wix design, so my instructions to Wix AI were not exactly complicated.
Keep the design.
Don’t change the existing pages.
Add the content I give you.
Use it exactly as written.
Apparently the word “exactly” leaves considerable room for artistic interpretation.
Entire service sections disappeared.
Buttons looked impressive but went nowhere.
The Resources page started reproducing.
And at one point Wix invented “GBPInternational”, which I still think deserves its own currency code.
Then we built a lead-capture form.
It looked excellent.
Fields.
Button.
Everything.
Except for the small technical detail that it didn’t actually submit anything.
Kimi described it as:
“The digital equivalent of a prop door in a movie set.”
Perfect description.
It looked like a door.
You could click the door.
You simply couldn’t go through the door.
My response to most of this was some variation of:
“lol, okay, baby steps.”
And eventually:
“Let’s try one more Wix AI roulette.”
There are moments in business where you should trust the evidence.
This was not one of them.
Eventually ChatGPT suggested something much more sensible.
Stop fighting the website builder.
Move the site to Netlify.
That turned out to be a much bigger moment than simply changing where the website was hosted.
Because suddenly changes could be controlled.
If something broke, we could roll it back.
If something worked, we could keep it.
And “deploy” slowly stopped sounding like something that might destroy the entire business before breakfast.
Then came GitHub.
I knew almost nothing about GitHub.
Branches, repositories, commits, pull requests...
Initially it sounded like somebody had deliberately invented a vocabulary to stop accountants from participating.
But the principle made immediate sense.
There should be one controlled source of truth.
Every meaningful change should be visible.
You should know what changed, why it changed and, importantly, how to undo it.
Which, when you think about it, is basically good financial control with worse terminology.
The website was no longer just a collection of pages.
It had structure.
And once the structure existed, something else happened.
The website started creating actual workflows.
People could complete assessments.
People could make enquiries.
People could request drafting work.
And somebody — unfortunately me — now had to keep track of all of that.
So Gmail stopped being just email.
We started structuring it around the business.
Executive Compass enquiries.
Drafting engagements.
Retainer work.
Lead generation.
Drafts.
Sent items.
Responses.
Follow-ups.
What had started with:
“Please make this button work”
had now become:
“How do we make sure that when somebody clicks this button, the right information ends up in the right place and somebody actually does something about it?”
Slightly different problem.
Then came automation.
Because once you have several service lines, lead-generation campaigns, incoming enquiries and follow-ups running at the same time, manually remembering everything becomes what I would describe as an unacceptable control environment.
So we started connecting the pieces.
Enquiries.
Email.
Follow-ups.
Assessments.
Client information.
Workflows.
And then the next question became obvious.
Where does all the actual information live?
An assessment cannot just exist inside an email.
A client cannot just be somebody I vaguely remember speaking to last Tuesday.
A drafting engagement needs a status.
A retainer needs a structure.
A completed assessment needs a report.
And preferably we should know which one belongs to which client.
That brought us to Supabase.
And that is where the whole thing changed again.
Because now we were no longer thinking about a website with forms attached to it.
We were thinking about the architecture of the business itself.
Clients.
Assessments.
Responses.
Reports.
Drafting engagements.
Retainers.
Leads.
Follow-ups.
Different services, but one underlying system.
Then the questions became much more interesting.
What happens when someone starts an assessment?
Where does the information go?
What happens when they finish?
How is the report created?
What happens if something fails halfway through?
How do we avoid creating a report for an assessment that doesn’t actually exist?
How do we keep drafting work separate from assessment work without building twelve different systems that refuse to speak to one another?
At some point, I realised something quite important.
I wasn’t building a website anymore.
The website had simply become the front door.
Behind it, Executive Compass was turning into a system.
And AI was involved in almost every part of that journey.
ChatGPT helping me work through the business model and eventually suggesting Netlify.
Kimi inspecting things and occasionally informing me, with admirable efficiency, that something was completely broken.
Copilot appearing periodically with another opinion.
Coding agents working against GitHub.
Supabase holding the structured information behind the operation.
Gmail becoming part communication channel, part workflow system.
And somewhere in the middle of all of that, Executive Compass itself started behaving exactly like ATLAS.
Make a claim.
Test it.
Look for evidence.
Find the boundary.
Try to break it.
See what fails.
Change the rule.
Try again.
The technology became a live experiment in the methodology.
Some things worked beautifully.
Some things failed spectacularly.
Some things required several attempts.
Some things required language I probably shouldn’t repeat in a corporate presentation.
But every failure left evidence.
And once you treat failure as evidence rather than a personal insult, it becomes surprisingly useful.
That, for me, is the real origin story of Executive Compass.
It wasn’t invented in one moment.
There was no grand unveiling.
It evolved.
A business plan became a research project.
The research project became ATLAS.
ATLAS started influencing the way the business itself was being built.
A website became a platform.
The platform needed workflows.
The workflows needed data.
The data needed structure.
And slowly, all of those pieces started connecting.
That is also why I don’t see AI as the product.
AI is extraordinarily useful.
It can research.
Challenge.
Write.
Analyse.
Build.
Code.
Test.
And, apparently, occasionally create a very attractive button that does absolutely nothing.
But the valuable part is still judgement.
What question did you ask?
What information did you trust?
What did you ignore?
What assumptions were made?
What would change your mind?
And when the system tells you that your clever idea doesn’t actually work, are you prepared to change it?
Executive Compass exists to make that judgement clearer, more visible and more testable.
Not to pretend uncertainty disappears because we added artificial intelligence.
And definitely not because the website finally has working buttons.
This has been independently funded by me.
There are no sponsors.
There are, however, definitely offers available now.
So if you’re a CEO, CFO, COO, or any other member of the corporate alphabet soup, you’re very welcome to have a look at executivecompass.global.
There’s paid stuff.
There’s free stuff.
And everybody loves free stuff.
So at the very least, go and take something before I change my mind.
Some of the offers might even be useful.
And, as an added bonus, the buttons work now.
Executive Compass.
Clarity.
Direction.
Better decisions.
Independently funded.
No sponsors.
Working buttons.
Mostly.
And yes...
the 13-Week Cash Flow Template is still missing in action.