Twelve platforms, three hours

Wed, 16 Sep 2026 00:00:00 GMT

I was about to spend ten days automating posts to twelve platforms.

The estimate was honest. LinkedIn needs manual re-authorization every 60 days: refresh tokens go only to approved partners. X rotates its refresh token, so if you don’t persist the new one on every refresh, everything silently stops. Bluesky wants link offsets in UTF-8 bytes rather than characters, which is its own kind of fun once Cyrillic enters the picture.

Seven to ten days of work before I’d trust it on a cron job.

Those are ten days I’d have nothing to post about. The tool for making content would eat the time I had for content.

So I built the boring version instead: a skill that splits one draft into per-platform files. Three hours. I post them by hand.

In two weeks I’ll look at where I actually posted, and automate only those. My guess is that three of the twelve survive.

Twelve products in a year: how I kill ideas

Wed, 16 Sep 2026 00:00:00 GMT

I have evenings and weekends for this challenge: ten to twelve hours a week. Over a year that is about 550 hours. Split across twelve products, each one gets under fifty hours, and that includes the idea, the build, the landing page and whatever attempt I make at telling people it exists.

Where the number comes from

I am Alexander. I have been writing backends in Python and C# for over ten years, all of it employed. I tried freelancing and it did not stick: the same work for a client, except now there are several of them and they pay less.

I wanted something of my own. You cannot guess demand in advance, you can only test it. Spending a year on one idea and being wrong costs a year.

While reading about micro-SaaS I came across Pieter Levels, who set himself twelve startups in twelve months back in 2014. Launching has got an order of magnitude cheaper since, and with AI on hand the number looks reachable. I like challenges, so I am in.

To be upfront: I have never launched a product and I have no ad budget. I plan to talk about the products on social platforms and niche communities. I am keeping my job. This site is the log: what I built, what broke, what the numbers were.

Hard filters: one “no” and the idea is out

Binary questions. They cut ideas that do not fit “one person, fifty hours, zero ad spend”.

1. The MVP fits in 50 hours. Three screens and one button, and that is enough for someone to pay. If the first version needs a month, the idea is fine but not for this challenge.

2. Useful to the first user on day one. He arrives alone, with no accumulated content and no community around the product. So anything two-sided drops out here: marketplaces, job boards, “find a specialist” services. You cannot launch those alone and unfunded.

3. The audience already pays for tools. Right now, to someone, for something similar: small businesses, freelancers, agencies, developers with a company card. An audience with no habit of paying will not acquire one for my product.

4. There is a place where the audience is already gathered. A subreddit, a chat, a forum, something with a name. Without an ad budget, distribution means going where people already sit. If no such place exists, there is nobody to show the product to.

5. No one else’s money and no one else’s documents. Payments inside the product, healthcare, passports, HR records: compliance and support eat more time than the product itself.

6. Cost per user close to zero. No heavy models on my own hardware and no manual steps inside. Otherwise every free user becomes a line item.

7. No single external dependency. If the whole product is a wrapper around one API, its owner changes the pricing or the rules whenever they like, or closes the door altogether.

Soft criteria: comparing the survivors

When three or four ideas make it through the hard filters, I pick with these.

Explains itself in one sentence. If the landing page needs an intro paragraph, the product is unclear, and I will be explaining it to every user in person.
One pain, minimum features. The product does one thing. Anything that can be cut from the first version gets cut.
I can find ten users myself in a week. That means a list of ten actual people I will message. No list, no build.
I still care in six months. Support outlasts launch. A product that bored me in week two will not get fixed in month three.

What is deliberately missing

No “market size”. At my scale the difference between a hundred-million market and a five-million one means nothing: the first hundred customers come by hand either way.

No “no competitors”. Missing competitors usually means missing demand. Competitors with paying customers confirm that people pay for this.

What is next

These seven filters are a hypothesis, not a method. They were written before the first launch and some of them will turn out to be nonsense. After two or three products I will come back and rewrite the list, noting which filter actually caught something and which one let through an idea I should not have taken.

Next post is about where ideas come from: where I looked, how many each source produced and how good they were.

If your list has a filter mine does not, tell me. Right now I am most interested in what I have missed.