Welcome to the latest edition of Empire of One.

Today's edition is about picking one business idea from a pile of a hundred, and using AI to make that choice stick. By the end of this issue, you'll have a repeatable method for choosing an idea, plus a 14-day plan for testing it.

You know that list of ideas in your Notes app: 40, 80, maybe 200 lines. Half are headlines you saved at 11pm. None have seen a paying customer.

Tonight you almost start building one. Then a video suggests 3 more, and you add those too. You close the laptop feeling informed, but you didn't get closer to your first $1,000. You only made the list longer.

Inside this issue:

The Real Problem

You've probably got enough business ideas to keep yourself busy for the next five years.

An affiliate website. A digital product. Maybe a small AI tool that solves a problem you've run into at work.

Every time you sit down to choose one, another promising opportunity appears.

I understand the temptation. Research feels productive, especially when you're learning something interesting.

But there's a cost to keeping every option open.

You can spend six months comparing business models without finding out whether a single customer will pay for what you're proposing.

And when you've got a full-time job and maybe an hour or two available after dinner, six months is a considerable investment.

I'd rather see you choose one reasonable idea, give it two weeks of focused attention, and find out what the market thinks.

You might make your first sale. You might learn that your original assumptions were wrong.

Either result gives you something useful to work with.

Pick one. Archive the rest. Put a date on the calendar.

Why Another Idea Never Resolves the Pile

When the pile keeps growing, the instinct is always the same: read one more article, watch one more roundup video, ask an AI one more time whether an idea could work.

None of that tests anything. You can spend weeks comparing options and still not know whether a single person would pay for any of them.

An idea that feels exciting this week can lose that shine by next week. Excitement is a mood, not a signal. The only thing that actually validates an offer is a real person deciding to pay for it.

So instead of asking which idea is "best," ask a narrower question: which idea on your list could survive contact with a real customer, inside a 14-day window you'll actually enforce?

That's not a creative exercise. It's a scoring job, mechanical and a little boring, and one of the few real advantages you have over everyone else still waiting to feel ready.

Score the Pile Like an Operator, Not a Collector

Stop asking which idea is "best."

"Best" is a vanity metric. It belongs to theorists who want a guarantee before they take a single step into the market.

If you want to operate with discipline, you replace that question immediately:

Which single idea allows me to run a clean, contained experiment over the next 14 days-using the exact hours I have, the skills I already own, and a downside I can comfortably absorb?

"Best" is an illusion.

Fit-for-the-next-two-weeks is an operational standard.

Here is how you handle the triage:

Do not attempt to score fifty or a hundred ideas. Scoring a massive list is just another way to stay busy without taking action.

You score eight to twelve. That is the limit.

If your notes app is currently an overgrown catalogue of ideas, cluster them before you do anything else.

Paste the raw list into a single document. Hand it to an AI with one clear directive: group every near-duplicate into a single category.

"AI newsletter," "ChatGPT tips," and "AI beginner guide" are not three different businesses.

They are one idea wearing three different outfits.

Collapse them into one entry.

Select 8 to 12 distinct concepts.

Archive the rest out of sight. You can always revisit them later, but for now, they are off the table.

Score each idea across five categories on a 1 to 5 scale:

Here are the five criteria:

1. Customer Access. Can you name a person, or a place those people already gather, that you can reach in 14 days without a following or an ad budget? 1 means invent an audience from zero. 5 means you already talk to these people.

2. Speed to First Test. How fast can a stranger see a real offer? A one-page outline plus a direct conversation is a 5. A custom app or a YouTube channel requiring 50 videos is a 1.

3. Demand Evidence. What exists outside your head? People already paying for a clumsy version. Repeated complaints in a forum you actually read. A tool coworkers ask you to explain. A 5 is specific and recent. A 1 is "I bet people would want this." AI's enthusiasm isn't evidence.

4. Current Skill Fit. Does this use something you can already do at a level a stranger would trust? A 5 is "I've done this for years, even if only as an employee." A 1 is "I'd have to become a different person by Friday."

5. Downside Protection. A 5 is cheap to be wrong: time, a simple page, a few direct messages. A 1 is a new domain, custom branding, three software subscriptions, and quitting your job.

Add the five numbers. Score out of 25.

Then apply two filters the numbers can't calculate for you.

Filter A: willingness to look average for two weeks. Infatuation fades when the first message goes unanswered. Willingness survives an unanswered message.

Filter B: pick from the top three scores, not from the romance. If your favorite sits at 11 and a less glamorous idea sits at 22, the 22 is the operator pick. Later is allowed. Later isn't this sprint.

The Worked Example: Teardowns vs. The Newsletter Trap

Let's compare two common ideas side by side. Idea A assumes you're already active in relevant communities and can ship a paid deliverable manually. Idea B assumes a newsletter needs ten back issues, branding, and months of growth before you can test demand — a leaner version, tested with a simple signup page and direct outreach, would score higher.

Idea A: active in two e-commerce communities, can record a review on Loom tomorrow evening, and store owners already ask publicly why visitors bounce. Idea B, scoped as a traditional newsletter launch: no existing design audience yet, and the ten-back-issue assumption pushes first contact with a buyer out by months.

Idea A produces real conversations and payment potential this week. Idea B, as scoped above, demands months of quiet work before anyone sees an offer, though a leaner version could close some of that gap. Idea A is the freeze for now.

The scores name the idea that produces evidence soonest, with the least setup, using skills you already have. Proof is allowed to be unglamorous.

AI handles the clerk work: clustering, restating, and scoring. Human judgment selects the winner.

Run the 14-Day Freeze on One Idea

Tonight or tomorrow night, run the scoring session. When a winner is on the page, you freeze.

The 14-Day Action Schedule:

The Three Freeze Rules:

The 14-Day Operator Checklist:

What counts as finished: one specific offer existed in public, a few real people could say yes or no, and you recorded their response. Silence is evidence too.

If the 14 days feel thin, ask one question: did the idea fail, or did the test never leave the house? If it never left the house, run the same idea with a stricter definition of shipped. If it left the house and nobody cared, promote the runner-up.

The Digital Leverage

Imagine sitting down on Saturday morning with 80 scattered notes about potential business ideas.

You've got half-finished research, screenshots, bookmarked tools, and a few ideas that looked promising at 11 p.m. on Tuesday.

Now you've got to sort through everything, group similar ideas, compare the opportunities, and score each one against the same criteria.

There goes your weekend.

This is the kind of work where AI can save you some serious time. Give it your notes, a clear scoring framework, and instructions to challenge every unsupported assumption. What might take you several hours manually could become a focused 20-minute session, although you'll still need to check its conclusions.

And here's where I think AI becomes particularly useful.

It can challenge the scores you've given your favorite ideas.

For example, you might rate an opportunity 5 out of 5 for demand because you're convinced people need it.

AI might come back with something like this:

"You've rated Demand Evidence a 5, but you haven't identified an active forum discussing the problem, named an existing product people are buying, or identified a single potential customer. Based on the evidence you've provided, this deserves a 1. Find some proof before raising the score."

That's a useful challenge.

Of course, AI can miss things, too. It can overlook relevant evidence or confidently question something you've already verified. You still have to check its work.

But having something challenge your assumptions before you spend two weeks building around them? I'll take that.

The bigger benefit is what happens to the time you've saved.

Think about the 90 minutes you might otherwise spend watching another YouTube roundup of promising online business models.

You could use that same 90 minutes to contact potential customers, publish an affiliate review, or put a simple $9 product in front of people who might actually buy it.

At some point, you've got to give the market a chance to respond.

And I'd keep the operation simple. AI makes it surprisingly easy to build an elaborate system for managing work you haven't started yet.

Here's how I'd divide the responsibilities.

Keep Human

You make the final decision about which opportunity to pursue and commit to working on it for 14 days without switching to another idea.

You handle the actual conversations with potential customers. You also decide what the market's response means.

If nobody responds, you'll need to determine whether the proposition is weak, you're reaching the wrong people, or you simply haven't done enough outreach.

Those decisions deserve your attention.

AI Assist

Let AI handle the work that benefits from speed and consistent organization.

Have it group your messy notes, turn vague ideas into clear, single-sentence business models, and audit each opportunity against your five scoring categories.

It can also draft your initial outreach messages, which you'll review and personalize before sending.

The point is to spend less time organizing your research and more time testing whether you've found a real opportunity.

AI Delegate

I'd hold off on autonomous AI agents until you've had verified, direct contact with potential buyers.

Once you've got evidence that people want what you're proposing, you can consider handing off repetitive parts of the workflow.

There's little reason to build an autonomous operation around an idea that hasn't survived its first customer conversation.

Automate

Nothing during the first week.

Seriously.

You probably don't need another workflow, integration, or subscription while you're still trying to establish whether someone will pay for your idea.

Automation can wait until you've identified a repetitive task worth automating.

Eliminate

For the next 14 days, stop collecting business ideas. You've already got plenty in the pile — give one reasonable opportunity a fair test before adding anything else to it.

Maybe you land your first sale. Maybe you learn your original assumptions were wrong. Either way, pay attention to that first verified signal: a real transaction, even a small one, tells you someone was willing to act, and that's worth more than another month of research.

The Copy-and-Paste Scoring Prompt

Paste the following directive into Claude or ChatGPT, attach your raw inventory of concepts at the bottom, and execute the run:

Act as a skeptical operations director helping me select one business idea to test over the next 14 days. I have limited evening hours and need to reach my first $1,000 online.
Here is my context:
- Available weekly hours: [e.g., 6 hours per week]
- Current professional skills: [e.g., copywriting, basic web design, B2B sales]
- Audiences or groups I already have access to: [e.g., local contractors, Shopify Discord, LinkedIn network]

Review my idea list below. Cluster duplicates, restate each distinct idea in one sentence (who it serves, the core problem, how money moves), and score each idea from 1 to 5 across these five categories:

1. Customer Access (existing direct reach vs starting from absolute zero)
2. Speed to First Test (ability to put a live offer in front of a buyer within 14 days)
3. Demand Evidence (documented proof that people currently pay for this solution today)
4. Current Skill Fit (relies entirely on competencies I already possess right now)
5. Downside Protection (inexpensive to test, with minimal cost if it yields zero response)

For every score, label the evidence behind it as Verified (something I've confirmed), Claimed (something I believe but haven't confirmed), or Assumed (a guess). If you don't have enough information to score a category, mark it Unknown instead of guessing. For any score above 3, demand one specific piece of verified evidence from my background or the market before accepting it. Rank the top three options and define the smallest viable 14-day test version for each. Explicitly flag any concept that quietly requires building an audience before making an offer.

My idea list:
[PASTE YOUR RAW LIST HERE]

If the model begins offering unearned praise or telling you how brilliant your concepts look, correct its posture immediately:

"Stop selling me these ideas. Stress-test them against the criteria."

Once the rankings are printed, review the data, make the executive call, and close the intake door.

Which idea will you freeze on for the next 14 days, and who will see the offer first? Reply to this email with that one sentence.

Whatever you pick, the next two weeks are simple: one offer, in front of real people, with their response recorded.