My first hire isn't allowed to decide anything

Every Monday I'm going to take one role from my one-person team and show you how it works: what it does, how it's set up, something it actually produced, what broke, and a template you can copy. I'm starting where I'd tell anyone to start, at the bottom of the chart.

The first hire is the scout. It works for my show, The Operator Economy, where each episode is a business one person could run. Twice a week it goes looking for problems someone would pay to have solved, and it keeps a short list of the ones that might be worth an episode.

It reads public conversation: search results, forums, professional communities, and what's changing in markets, rules and prices. When something looks promising, it writes it up as a possible business. Who'd pay, what the problem costs them, what someone could offer, why now, and the question most likely to kill the idea. Then it checks the idea against everything I've already covered or turned down, because a renamed duplicate is still a duplicate.

What it can't do is the more important half. It can't turn an idea into an episode, put it in the queue, write a word of script, contact anyone or spend anything beyond a small, capped search budget. And it isn't allowed to treat a company's own marketing as proof that buyers exist.

What it actually did

On October 11 it looked at two ideas and turned both down.

One was helping small trade subcontractors get through the safety paperwork that big industrial clients require before they'll hire them. The need looked real. Small firms describe paying consultants a lot of money to get through it. But a tool already sells AI-drafted answers for exactly that paperwork, directly to the subcontractors, for far less than a consultant. Whatever's left for a one-person service is what that tool and the owner can already do together. So it went in the rejected pile, with a note on what would reopen it: a recent account of those AI-drafted answers failing review while firms pay someone to fix them.

The part of the run I like best is the boring part at the top. It lists every place it looked, which searches came back empty and which sites refused it. A shortlist with no record of where it looked is a list of opinions.

What broke

On September 21 the scout lost its best source. The run the day before had quoted forum threads where people described their problems in their own words, which is exactly what you want. The next morning every route to that forum was blocked.

The tempting fix was to find a way around the block. I didn't. For almost a month the scout read that forum only through what search results show, and wrote the block down as a limit of each run. It never counted the silence as evidence that nobody was talking about the problem. This week it got a paid search tool, with a small cap per run, that sometimes shows more of a thread than a results page does.

The other early break was mine. My first episode's idea came with a great headline number that didn't survive the first honest read, because it was published by firms that sell the service. Now every number in a lead says who published it, and a seller's marketing is context, never proof.

Copy this

ROLE: [name, e.g. Scout]

JOB (one sentence):
[What it does, for whom, how often.]

INPUTS (the only places it may read):
- [source 1]
- [source 2]

SCHEDULE:
[e.g. Mondays and Thursdays, 7:00am]

OUTPUT (same shape every time):
- Title
- Why it matters, in one line
- Source link, and who published it

LIMITS (it may never):
- Spend money or buy anything
- Contact anyone or send anything in my name
- Promote, approve or decide anything
- Change anything it didn't make

APPROVALS:
- Everything it produces waits in [where you'll see it each day] until I say yes.
- If it isn't sure, it says so and leaves the item for me.

Start slower than I did. Twice a week is plenty while you work out what you want from it, and the review takes five minutes: open one source, cross out anything you wouldn't put your name to, and move at most one item forward.

Shipped recently

• The Operator Economy's new episode, on the one-person business that gets small hotels booking direct, premieres today on YouTube.
• Little Fables now says plainly how long it keeps what a parent tells it, and after the book arrives, the parents decide.
• My assistant can now sort the websites it uses into ones it can only look at and ones where it has my standing OK to order, up to a limit I set.

Next issue, next Monday: the research bench, the role that checks the scout's work. If you'd rather get these by email once the list starts sending, you can sign up at mpthaker.xyz.