Will AI make me redundant?

What happens to our jobs when turning an idea into a working software product takes an afternoon instead of a quarter? Will we still need engineers and delivery leads when we don’t need teams of people to build software?

Jevons’ paradox.

When steam engines got more efficient, we didn’t use less coal. We built more engines. When servers could host more applications with virtualisation, we didn’t buy fewer servers, we purchased more servers because they unlocked more value.

If building software gets cheaper, we’ll end up with more software. But we’ll also have more developers. Developers are the steam-engines and servers in this analogy. Being a developer will become more accessible, and these new developers will build weirder systems, serving smaller and more specific needs.

Programmers are going the way of the typing pool. The lawyers who used to dictate letters to their secretaries retired. Then the next generation of lawyers just typed and emailed without help because it was faster to DIY. The skill gets absorbed into the person who needs it. Shifted-left in DevOps speak.

The risk project manager who’s spent twenty years building mental models of a specific domain can now turn that knowledge into software. Specialist expertise imported into organisations from expensive consultants becomes self-serve, best of breed software tools.

We end up with more products. More niches. More integration points. The software landscape becomes larger and harder to navigate. And more people will be needed to figure out how all of it fits together.

Adoption remains hard.

While we can build software in an afternoon, getting people to use that software remains a challenge. Today’s vibe coded software is often built for one user. One context. One person’s idea, for themselves.

Getting other people to trust it, adopt it, build their work around it. Getting something into an organisation. Selection, architecture, configuration, integration, security reviews, change management. This stuff isn’t AI prompting. This is person-to-person relationship work that will still need to be done.

And enterprise environments are complex. Enterprise software is often monolithic, tangled up in numerous often undocumented workflows and integrations. Understanding the impact of a change in an enterprise environment requires analysis and domain knowledge. This is why enterprises get trapped with expensive, inflexible and out of date systems.

If AI makes integration genuinely cheap — if customisation becomes trivial, if stitching two systems together stops requiring a six-month project — then monolithic platforms lose their gravitational pull. Smaller, more specialised tools become viable.

Pressure mounts to integrate smaller best of breed software that is more tailored to specific needs, because the perceived cost of this will be lower. Organisations will need more people not less to help them navigate this larger software field and integration maze.

What to do about it?

From this perspective, all the jobs that deal with people, processes and change will become even more important. Product management, business analysis, change management and integration expertise becomes even more valuable.

Core business domain expertise will remain valuable but these roles will also require AI literacy. Being able to build software will become baseline skills, that you wouldn’t even mention on your resume, like ‘proficiency in word processing’.

Following this logic, businesses will need more people, not less people. Delivery leads may have a future; supporting project teams of product managers, BAs and business SMEs. And engineers will either step into into these roles, or will build AI platforms that enable users to safely build and integrate their software within their enterprise.

One thing I personally hope for; with decreased cycle time, we’ll be able to iterate faster, run more experiments and stay closer to the people we’re building for. Corporates could become more agile and innovative. This sounds more fun and fulfilling.

I am loving bringing my software ideas to life with less effort. But in writing this article I have realised the real work of the future will be go-to-market and iterating with users. I need to level up. It’s not enough to build for myself anymore.