TL;DR:
- AI tools make app building much more accessible.
- Apps still need secure, maintainable infrastructure.
- Only standardized deployment makes citizen-built software scalable.
Building apps is easier now
The days when only a few, very well educated software developers could build apps are gone. With tools like Claude Code, Codex, or GitHub Copilot, it has become much easier to do so. While one still has to steer AI (Outsourcing thinking, not understanding), manual code generation is mostly unnecessary nowadays. This gives a wide range of people the opportunity to build services that they think are needed for the environment they are operating in. This makes a lot of sense, because they are the ones who know best what they need.
This may be the end of low code as we know it, in my opinion. For most use cases, it no longer makes sense for people to click together predefined modules to create services that never fully meet their requirements. Too many use cases need to break out of the very limited options provided in low code environments.
As it has gotten so much easier to directly create code, we have become much more flexible. While people had to explain their needs to professional software developers before, this is now no longer necessary. They can build themselves, iterate much faster, and have only a very thin AI layer between what they need and what is actually being built. We can be so much more innovative this way.
But do apps have a home?
But those apps need to run somewhere in a secure and reliable way. The data processed in the background needs to be stored somewhere. The background services they use need to be hosted somewhere. The code needs to be stored somewhere. The app needs to be maintainable. Otherwise, services built by citizen developers will not be successful in the long run, as bugs cannot be patched and dependencies cannot be updated seamlessly.
What makes this sustainable?
What can we do to make sure that apps built by anyone can be deployed in a long-term, secure, reliable, and scalable way so that they can actually bring added value to their users?
Build a unified environment
We need an environment where different apps can be deployed in a standardized way. This includes a deployment infrastructure that is already set up, where apps can be added and deleted as needed. It also includes databases and AI models that are readily available and can be used by the apps as needed. We also need a standardized way to monitor and update the apps to track usage, add improvements, and fix bugs over time.
Inspired by the frontier AI lab Anthropic (Lenny’s podcast interview with Cat Wu), we have to make it as easy as possible to create and deploy new apps. Only by giving people the option to use new applications can we find out if they add value and are actually useful. At Anthropic, this low-process environment allows engineers to create apps and ship them in a preview version in a matter of days. This is how Claude Code and Claude Cowork were created.
With tech-savvy people distributed across whole companies, organizations could then be transformed from within by the people who know their job best. All people would need is a bit of technical understanding, a coding environment, and access to AI tokens.