We've spent the last three days building an AI coach from scratch. Part 1 was the vision. Part 2 was the morning check-in. Part 3 was the daily review that doesn't lie.
Now it's time to put the coach to work on actual decisions.
I had three real business opportunities come across my desk last week. I ran all three through my AI coach. The scores surprised me. What I did with those scores surprised me more.
Opportunity 1: The Side Project That Shouldn't Make Sense
Last week a friend and writing mentor told me about a new payment protocol called x402. Here's the short version: the internet has always had an HTTP status code called 402 — "Payment Required." It's been sitting there unused since the web was invented. Nobody knew what to do with it. Now Coinbase, Cloudflare, and a growing foundation have built an open protocol on top of it that lets any website or API charge tiny amounts — fractions of a cent — instantly, with no accounts, no credit cards, no middlemen. Stablecoin payments over the internet, as simple as loading a webpage.
Why does this matter? Think about a writer who wants to charge 10 cents for an article instead of forcing readers into a $10 monthly subscription. Or a small nonprofit that wants to accept a $2 donation without paying $1.50 in processing fees. Or an AI agent that needs to pay for data on behalf of its user. x402 makes all of that possible. Since launching, the protocol has processed over 35 million transactions. Stripe added support in February. This isn't theoretical.
Around the same time, I'd been listening to a Y Combinator podcast where YC partners were talking about the next wave of startups. The message was clear: the best founders in their latest cohorts — which are now roughly 60% AI companies — are building software for AI agents, not humans. The application layer is shifting. SaaS as we know it is becoming invisible, replaced by agent-native infrastructure. If YC's best applicants are thinking 6 months out to build for agents, that's something worth remembering. Maybe even investigating.
Sunday morning, I started connecting these two threads. What if there was a way to use x402 to solve a real problem I'd been banging my head against for months — I'll get more in to that problem once I have something to show. So I bounced between Perplexity and Claude, built a rough plan, and ran it through the coach.
The best part about this exercise is that the coach said my ideas were off, bad fit, but did point out an example for a prestigious grant for $150,000 could be the best fit. Work on this for the grant application but have reasons for doing this if I didn't get the grant. A full backup plan if that failed.
Here's how the coach scored it across five dimensions, each on a 1-to-5 scale:
Mission alignment: WEF, powered by Finn 5. This was directly connected to the work I care about most. Not a distraction. Foundation-level infrastructure.
Content value: 5. Three weeks of daily posts write themselves. The decision, the research, the build, the application process. Real BIP content that people would actually want to follow.
Prospecting value: 5. Every youth empowerment organisation I'd bring into this project is a conversation I wanted to have anyway. A warm intro, not a cold email.
Downside risk: 5. About 35 hours of work. Around $50 in hard costs. If it doesn't work out, I keep the tool, the relationships, and the content. Minimal downside, asymmetric upside.
Revenue timing: 2. This was the only weak spot. The project doesn't generate client revenue directly for months. Every hour spent here is an hour not spent closing deals. The prospecting angle offsets this partially, but it's real.
However, what the AI coach didn't know was that I was already sitting on what I knew was a verbal for revenues coming in the next 30 days and prospecting was no longer the highest priority. I had some cycles until those revenues would be coming in.
Total: 22 out of 25. On paper, that's a strong go. But the revenue gap is a legitimate concern when you're in month one with zero recurring income, but the $150,000 grant win could offset that. Not in the bag, but something there.
So I went back to something I've been leaning on more and more. The vibe.
Sorry if that sounds too surfer dude, but I can't find a better word for it. Call it gut instinct if you want. That's maybe close, but vibe is the pure energy that emanates from something; a person, a situation, a project, an instinct, anything. It gives energy, you know this but can't put your finger on it. That's a vibe, and the vibe on this project was a 5. I wanted to build it. It would directly benefit youth organisations. And as a worst case, I'd have a way to collect small donations for the Finn Wardman World Explorer Fund — something I'd already spent months trying to solve. The WEF is based out of Bermuda, where the banks use technology from the 1980s. Collecting small international donations has been a genuine pain point, not a theoretical one.
The vibe pushed me over. I started building.
Opportunity 2: The Enterprise Dream
I ran this one through the coach and it scored 23 out of 25. Nearly perfect across every dimension. Mission-aligned, revenue-generating, content-rich, low risk, and immediate.
The vibe? A 9 or 10. An entrepreneur's dream scenario. Being let in to a big opportunity with a clean slate to design and build world-class AI layers with no committee telling me what I needed to do.
That's a definite yes.
Opportunity 3: The Safe Bet That Wasn't
A smaller client opportunity. Reduced rate. Pile of work. An expectation of getting deeper behind their tech stack, more like an employee than a fractional operator.
The coach scored it 12 out of 25. Below threshold on nearly every dimension that matters.
The vibe was weak. The funds barely moved the needle and the opportunity path seemed bare. I might take it to increase revenues at such an early stage, and maybe it leads to some lessons or I could hire someone good at AI automations to handle it. But the energy wasn't there.
The Two-Filter System
So this is how I'm using the AI coach now. Every opportunity gets two scores.
The coach gives me the analytical breakdown. Five dimensions, each scored 1 to 5. Twenty or above is a go. Fifteen to nineteen, proceed with caution and name the tradeoff. Below fifteen, walk away unless something else overrides it.
Then I check the vibe. And the vibe isn't some mystical thing. It's the accumulated signal from everything I know about myself, the people involved, the work itself, and whether I'd be excited to wake up and do it tomorrow morning. It's the thing that made me quit a high-paying job three weeks ago. It's the thing that made me cut loose a co-founder role the next day. It hasn't been wrong yet.
The coach gives you the numbers. The vibe gives you the signal. Together, they're the best decision system I've ever used. Maybe I need to give this coach a name. I'll noodle on that. But it's becoming a trusted advisor. It's not charging me for advice. It's giving me advice I actually trust, and I can use it for nearly every decision I make.
Between the analytical scoring and the vibe instinct, I'm about to step into a world-class enterprise opportunity and I've started on a project that could land a $150,000 grant. From where I was less than 30 days ago, I'm in a different world. A total reinvention, and feeling froth for work.
Even though I had a verbal on revenues, like a good salesperson, I'm not counting those yet.
Revenue: $0 | Clients: 1 | Prospects: 4
Rather than prospecting, I'm now moving to fulfilment as the number one priority and telling my coach about the strategy shift. More on that tomorrow once we book revenues.
Day 11 of 365.
This is Part 4 of the AI Co-Pilot Series. Part 1: Design Your Perfect Day. Part 2: The Daily Check-In. Part 3: The Daily Review.