Things Munim has been saving about how websites are being built now, and how people are using AI to find customers. Newest at the top.
@the_cyw A builder who makes and sells small tools that get an AI to produce a finished website.
Three finished sites in the video, each made in a single go for about ten to fifteen dollars of computer time. The point is not the price. It is that layout, motion and imagery arrive together instead of being assembled one piece at a time.
@the_cyw The same builder as above.
His own admission that these scroll-through sites look wonderful on a laptop and fall apart on a phone, and what he changed. A useful corrective to everything else in this section.
@yigitakinkaya A Turkish developer who takes apart techniques other people are showing off.
A site that feels like moving through a real space, with no three-dimensional graphics anywhere in it. It is one short video, played forwards and backwards as you scroll. This normally needs a specialist and several weeks. Here it is an afternoon. Written in Turkish, so use your browser's translate.
@ClaudeDevs The official account of Anthropic, the company that makes Claude, the assistant Munim builds with.
A shared page has always shown everyone the same frozen contents. These can go and fetch something current, for each reader. It moves a page a long way towards being a small application.
@titepommee A designer who posts about working alongside these tools.
The closest one to your own work. The claim is that the limit is not what the tool can do, it is your vocabulary: if you cannot name the pattern you want, you cannot ask for it. You name things for a living, so your view on this is worth more than his.
@marlowxbt Writes short breakdowns of how small products grew.
The best single thing here. Growth was not advertising, it was built into the product: one host pulls two hundred guests in over a single evening, and some of those guests later hold their own events. A guest at someone else's wedding becomes next year's host.
@levelsio Pieter Levels, a Dutch maker known for building and running several profitable products entirely on his own.
He counted three paid products rebuilt and released free in one day. His open question is whether ordinary people will actually switch, and what the paid companies do then. It bears on anything you and Munim intend to charge for.
@dremnik A developer who rebuilds paid software as free alternatives and publishes the result.
The live example of the point above, by someone doing it deliberately and repeatedly. Read it straight after the previous one.
@NathanGeckler Publishes teardowns of how apps introduce themselves to a first-time user.
Forty thousand downloads in a month. What he praises is what it leaves out: no long questionnaire, no invented personal plan, and it waits until your second visit before asking for money. A short lesson in restraint.
@gregisenberg An American founder who publishes startup ideas to a large audience and runs a podcast about them.
Start here if you read only one. Not theory. Each item is a small task a person used to do, connected to the information it needs and left running. One example: watch for a customer whose usage doubles in a month, and put a human in front of them.
@gregisenberg The same founder as above.
Two are worth your time. Being the answer an AI assistant quotes is becoming the new version of ranking on Google. And when anyone can build the product, the audience becomes the part that is hard to copy, so you build the following first and the product second.
@femke_plantinga Writes about running day-to-day marketing with these tools.
The clearest short explanation in the list. Either you set the goal and let it find its own route, or you draw the route first and let it fill in each step. Her rule: the first for one-off work where no path exists yet, the second for repeatable work you already understand.
@jacob_posel Writes about how companies reorganise themselves around these tools.
The company-shaped version rather than the task-shaped one. The central idea is one shared store of what the company knows, its documents, notes and decisions, that every tool reaches into instead of each one starting from nothing.
These were picked out of a much larger pile, and most of that pile was thrown away. A lot of what circulates on this subject is written to be shared rather than to be true. One post claimed a study had proved five AI assistants could replace a marketing team of fourteen people and save four and a half million a year. No such study exists. Nine others were the same paragraph reworded by different accounts, two of them putting invented quotes in the mouths of real engineers.
So treat the numbers in anything here as the author's own, not as measured fact. The two you can lean on hardest are the wedding app and the sign-up teardown, because both describe something the writer actually watched happen rather than something they predict. The weakest are the lists of opportunities, which cost nothing to write and are never scored afterwards.
If a claim here matters to a decision you are about to make, say so and Munim will go and check that one properly.