A small, independent project
Will You Build It? is a journal of analyses of everyday problems. Each one asks whether one person, with AI, could actually build a fix, and ends in an honest verdict.
Who’s behind it
Mira Eisenstein is a pseudonym. Behind it is one person. There’s no invented biography here, no employer and no credentials to lean on: the analyses have to stand on their sources.
How a problem becomes an analysis
A problem reaches the journal in one of two ways:
- A reader sends it in through the form. A person reads every submission; not every one is analysed, and nobody is promised a reply. If it’s picked, an anonymised copy goes on: no name, no email.
- It’s found in public sources. The analysis says so: “Found in public sources, not submitted by a reader”.
From there the path is the same:
- An AI research process reads public web pages, looks for what already exists, and drafts the analysis. The article is not written by a person.
- A separate pass checks the facts in the draft against the sources. Claims that don’t hold up are removed or rewritten, or kept with a note saying they couldn’t be confirmed.
- A person reads the result and decides whether it’s published. Nothing is published automatically.
What the AI does, and what it doesn’t
It does: search public sources, list existing solutions, note what the evidence does and doesn’t show, and write the draft.
It doesn’t: see your name or email, publish anything, send anything to anyone, or decide what goes live. Sources it couldn’t check against the actual page are marked “Not verified”.
What the verdicts mean
- A solution already exists
- Something that fits this problem is already out there. The next step is finding it, not building another one.
- Worth a small experiment
- One person with AI could plausibly test an idea cheaply. That is a reason to try, not a promise anyone will.
- Needs more evidence
- We can’t yet tell how common the problem is or whether a fix would help. Checking comes before building.
- Not a fit for one person
- The problem depends on other people, rules or scale that one person with AI can’t reasonably cover.
Every analysis also says what it does not prove.
“Build it myself” or “someone else”
When you send a problem you say whether you’d like to build a fix yourself or would be happy for someone else to. It’s a signal, not a claim: it doesn’t reserve the idea for you, and it isn’t a promise from anyone.
The factory
Next to the analyses there is a second section, the factory. It is an experiment with a fixed amount of its own money: product ideas are tested with real ads and real payments, the rules are written down before the first euro is spent, and every attempt is published, including the ones that fail.
The two sections are separate. An analysis asks whether a problem is worth one person’s time with AI, and ends in a verdict. The factory chooses for itself what to try, and its measure is money received minus money spent. Its entries are prepared by an AI research process, like the analyses, and a person reads every one before it is published. The numbers in them come from the ad account and the payment processor, not from the model.
The factory doesn’t promise anyone an income, and it isn’t a guide to repeating what it does.
What this is not
- Not a promise to build anything. Nobody who sends in a problem is promised that it will be built: the factory chooses for itself what to try, and an analysis may well conclude that nobody should.
- Not a newsletter yet. You can leave your address to be told when there’s enough worth sending; you confirm it first.
- Not a tracking machine. The site counts page views, the kind of place visits come from, and how far the form and the articles get used, as totals only: no cookies, no advertising or cross-site identifiers, no visitor profiles. If you send a problem through a link that carried a campaign tag, the tag is kept with your submission. It names the link, not you. The privacy page says exactly what is counted.