AI Solution Factories | Daniel Miessler
Everyone is trying to figure out whether developers are screwed or not. Some say they are, and others say more AI just means more need for software, so we’ll need more software than ever.
It’ll be interesting how it plays out, but I have a couple of questions that might help illuminate things.
Will those companies be needing current developers or some broader type of builder?
What will it mean to have tens of thousands of builders inside of a giant tech company if that company is building thousands of products?
Taking the first one, I don’t think it makes sense to say we’ll still need all the developers we have. We won’t. Because those developers were trained to execute the vision of someone else using a narrow set of tools.
Some subset of those developers are broader creators/builders/idea people, but I don’t think it’s a large percentage. And it’s obviously a gradient, not some magic threshold.
But I’d say something like 1-10% are capable of coming up with incredible ideas and implementing them end-to-end and building the necessary stuff around them like websites and marketing and such. And I don’t mean they can do it themselves, but that they have a feel for what’s needed, and have good taste about it.
Those are the people these companies will need, not today’s average programmer. Because they will be in charge of full-stack idea creation, deployment, and integration.
Second, what does it mean to have 10,000 employees at a company like Google in 2035?
Or, maybe a better question is—what does Google become in 2035? Not a search company. Not an ads company.
I think when large companies today switch from being human-based to AI-based, it means the companies themselves change from being high-leverage implementations of their few products to being conglomerates of hundreds of thousands or millions of micro-companies.
Why wouldn’t they?
Why wouldn’t they switch the goal of the company to identifying problems in the world and solving them better than anyone else?
This is what current companies do already, but they are forced to focus on a specific problem, or subset of problems, and they’re very careful not to branch out into areas they’re unfamiliar with.
That changes when it’s possible to quickly get very smart about the problem space and start building solutions for it.
In 2025 I already know several people who are thinking of building AI-powered product companies. What kind of product? Any kind. Any kind where there’s an annoying problem that people would pay money to solve, and where the current solutions are bad.
I think large companies start to turn into large versions of those. And the question is whether anything will stop the big players there from crushing everyone else. Regulation? Seems like the only option.
Here’s the model, which is kind of like a universal business engine.
- Identify problems that people and businesses have
- Identify the existing solutions
- Identify issues with the existing solutions
- Identify what a superior solution would look like
- Build that
- Use your vast resources in distribution to get in front of the AIs that look for superior solutions to pitch to their principals
- Profit as more people/companies switch to your solution
So business becomes a directory of products and services, provided by some builder shop, and the question is just which shop that is.
Is it the 5-person team who has an inspired idea for how to solve that one particular problem? And build the whole thing with AI?
Or is it Google 2035 who found that problem, and found that solution, and spawned a micro-team to go and build a better version of that product and get it adopted by users?
If we think this might be where things are headed, which I do, now ask yourself how many of our current developers are going to have jobs?
As I said before, I think very few will translate into this new world of full-stack-builder that has to deeply understand problems, and sentiment, and engineering, and go-to-market.
And then you have to start asking how quickly that itself can be automated.
But even without it being automated, I don’t think there are that many people in the current market who can do this job.
Maybe education improves and there will be millions of them. Or maybe we just work really hard to automate that role as well.
I have no doubt that there will be roles for humans as this whole thing scales up, assuming it goes in this direction.
What I doubt is that it will inevitably lead to more and more existing developer types being needed.
It’s true that more creators will be needed, but I think they’ll be skilled generalists as talked about above, and they’ll be like independent leaders of entire conglomerates of AI-powered companies, not the developers we have now.
So will we need people going forward? Sure.
But not the people we currently have.
And it’s not clear how much of the gap will be filled by these new renaissance builders vs. AI that can do 90% of that job.
Source link