When everyone can build anything
Issue 233: What happens when the price of building software goes to zero?
During a recent commute in San Francisco, a new personal routine emerged. I took my usual Waymo route to the restaurant. I pulled out my phone as normal during the ride, but instead of scrolling social media or checking work email, I launched the Replit mobile app from my home screen dock. One of my 2025 goals is to build my own software before paying for any. I had an idea to migrate my Airtable base of design talent—my CRM as a spreadsheet. I launched the Replit Agent and copied the prompt I had written in Obsidian.
As I arrived at the venue for dinner, the first version of the app was shipped. Though it’s a simple CRUD app, it’s amazing to think how quickly functional software can be constructed. You deem it so, and AI gets to work.
As a former employee, I have a clear bias towards Replit. However, there are a lot of compelling players in the space that go from idea to app, fast: Vercel’s v0, Lovable, Bolt, Ohara, and more. This chapter of my career is focused on access, empowerment, and giving people superpowers—at Webflow, Replit, now at Atlassian and as an investor. Everyone should have access to building their ideas that can change the world.
However, as Syndrome, the antagonist from Brad Bird's The Incredibles says, "When everyone is super, no one will be." This quote begs the question, what happens when everyone can build everything, and the cost of shipping software goes to nearly zero? Let's explore the implications of the rapid-paced environment, how design’s value and role may shift, and the incentive structure on the other side.
The implications
AI is accelerating and it isn’t slowing down. Though I’m optimistic about this space, it’s an important reminder that moving fast requires controle and durability. If you’re riding a motorcycle at high speed and don’t have a clear handle on what you’re doing, bad things will happen. It’s important to address implications of what you’re creating as there is progress.
One implication I’m concerned of is the increase of AI slop—low-quality media created by Generative AI (which includes code). In defense of AI slop, humans can also create slop; not an exclusive issue to AI. This is a challenge in any category that grants access and empowerment to people making things they didn’t know how before. This happened with no-code site builders when people started publishing low quality and accessible sites. The solution isn’t gating the technology but creating a clear progression in comprehension for the end users.
The next problem is human adoption of AI. In any product you build, the largest competitive moat you will face will be existing human behaviors and what they use. This is why so many startups are still competing with Microsoft Excel or any travel planing app competes with a shared Google Doc or group text. If humans do not adopt the solution, none of this matters. AI is still in the early adopter phase but the general masses haven’t adopted in in their every day workflows. With things changing constantly, they may hold off to see how things normalize or not adopt it at all.
Design’s evolving role
The headlines and online discourse will tell you design is dead. There is no need to use any drawing tool or the AI slop created won’t require designers anymore. That’s far from the truth. The importance of design will remain, but the contrast will increase on what’s crucial in this next chapter. In a recent inverview, Amjad Masad talked about how engineers in the middle get squeezed out in favor of deep engineering and great product builders.
This will also happen to design. Companies will always have a need for a specialist such as an iconographer. Roles with deep expertise will be safe. If you’re not in that category, I recommend having a few areas you’re strong at in the T-shape and be a generalist who can cover more ground. As Masad says, run away from the middle that gets squeezed out!
Once the squeezing is complete, design’s role is going to be elevated in the age of AI. I think of the four types of software in the future: commercial, boutique, personal, and disposable. There is an opportunity to build better tooling and experiences for end users to participate in making software. The commercial space is essentially today's Enterprise SaaS landscape and presents the opportunity of bringing new innovations into a highly secure and regulated space. Personal software likely has end users with a different level of comprehension with creating software; resulting in a fine balance of abstraction for them.
Boutique software is where designers directly contribute to their craft. There is a class of software that doesn’t need to be boutique and served for utility purpose. I don’t spend my time pondering who will build the Arc Browser experience of paying my electric bill. When utility jobs to be done are accommodated, a class of boutique software will bloom for people who care about quality or have brand loyalty. This is similar to a person who might have an affinity towards a brand of apparel and find value in spending more for it the same way people might support Andy Allen's Not Boring Software.
The new value of design
When everyone can create everything, what will remain hilighted as valuable in design is the ability to create new originality, high craft, and brand experience. Let’s look at new originality. AI (and people who work in production roles) are good at reproducing everyday things based on existing information and data. What will be even harder is for new and original ideas to form. It’s an exciting challenge for a designer to invest something new against a wealth of information like AI.
In order for new originality to exist, the role of taste and curation are essential as new inputs for Generative AI. Taste may be a topic that feels fluffy to people, but it's a key input of what to make in the first place. Insights are another input that designers naturally bring into their research. When everything can be built, understanding what is important and why is at the front.
Finally, taste is most impactful when it can be paired with great execution of high craft. The ability to make something is an eternal skill needed in design. In addition to being excellent at the way of making today, those who have a high bar for craft must seek new ways of working with our emerging technology and new tooling.
The future impact
This is the worst our AI tools are going to be—we’re at the ENIAC phase. In addition to the cost of compute and usage going down, developer experience will flip from the majority of time spent is on writing code and shipping software to being more reliable and autonomouse. As a result, it’ll allow more time for humans with their sentient teammates to generate more ideas of what to build and why vs. how to.
Second, the tooling improve iterations of Generative AI to reduce slop. today’s tools go from A to Z with no breakpoint in between. The more control we have on our outputs, it’ll merit more predictable and higher quality results.
Finally, design’s impact will return to the craft of building and there will be a de-emphasis of strategy and planning. In one of my favorite talks, The Secret Truth About Executing Great Ideas, Frans Johansson says, "A strategy enables you to act." Most strategic initituives were either performative measures by people to show the value of their job or de-risking the time it cost to execute strategies. If the execution cost decreases dramatically as we speculate, then strategy shifts to more rapid insight to iterate.
Generative AI’s impact is similar to that of the 3D printer. The quality of the output depends on the input. Designers have the opportunity to invent new inputs that result in outputs the world has never seen before, and that’s exciting.
Hyperlinks + notes
A collection of references for this post, updates, and weekly reads.
Remote work is a blessing and a competitive edge by Adrian Mato
The Art of AI Reasoning: How to Think about Thinking Models by Natasha The Robot
Try the Replit Agent and get $10 in credits
This AI Tool Lets You Build Apps Faster Than Googling | Amjad Masad (Replit)
Check out Design Business Company’s post about how we partnered with Dinamo Typefaces to create Replit Diatype
Job opportunities
Medium is hiring a Principal Product Designer → A wonderful opportunity to work with Cassie McDaniel
As artificial intelligence eats everything:
1. EQ becomes much more important than IQ.
2. Taste becomes much more important than execution.
3. IRL experiences become much more important than URL (online) ones.
You wouldn't believe - i was literally finishing up my next post on the personal software topic. Really like the implications that you highlighted, David. Definitely going to refer this post in my piece. Happy sunday!