For decades, the software development lifecycle has remained remarkably consistent.

Gather requirements.

Design a solution.

Write the code.

Test it.

Deploy it.

Maintain it.

The tools have evolved, agile replaced waterfall for many teams and cloud transformed how software is delivered, but the underlying assumption has stayed the same: humans drive every stage of the process.

That assumption is beginning to change.

We're moving beyond AI as a coding assistant and towards AI becoming an active participant in software delivery. Industry researchers increasingly describe this as an Agentic Software Development Lifecycle (A-SDLC), where AI agents contribute across planning, implementation, testing and operations under human supervision.

That doesn't mean developers are disappearing. It means their role is evolving.

From writing code to directing outcomes

The first wave of AI development tools focused on productivity.

Autocomplete became code generation.

Code generation became automated testing.

Today, AI agents are capable of understanding repositories, implementing features, writing documentation, generating tests and opening pull requests with minimal prompting. Some can even monitor applications in production and suggest or implement fixes within defined guardrails.

The engineer hasn't become less important.

The work has changed.

Instead of spending hours writing boilerplate code, engineers are increasingly defining intent, reviewing outputs, making architectural decisions and deciding whether AI-generated solutions are appropriate.

The value is shifting from execution towards judgement.

The new lifecycle

Traditional software delivery looked something like this:

Requirements → Design → Development → Testing → Deployment → Maintenance

An emerging AI-enabled workflow looks more like this:

Requirements → AI-assisted planning → Agent execution → Human review → Automated testing → Continuous deployment → AI-assisted monitoring

The sequence is familiar.

Who performs each stage is what's changing.

The bottleneck is moving

For years, the slowest part of software delivery was writing code.

Today, code is becoming easier to generate.

The new constraints are different.

Clear requirements.

Reliable data.

Security.

Governance.

Architecture.

Human review.

Many organisations are discovering that AI can accelerate delivery dramatically, but only when the surrounding engineering practices are mature enough to support it. That's one reason many AI pilots struggle to reach production despite impressive demonstrations.

What does this mean for hiring?

It's tempting to assume businesses simply need more AI Engineers.

We're not convinced that's the full picture.

As software development becomes more agentic, organisations will need engineers who can design systems, review AI-generated work, understand architecture and make good technical decisions under uncertainty.

Those skills have always mattered.

They're becoming central to the role.

Technical ability will always be the foundation, but the engineers who create the most value may increasingly be those who combine technical depth with product thinking, systems design and sound engineering judgement.

The future isn't autonomous. It's collaborative.

The conversation around AI in software development often swings between two extremes.

Either AI will replace engineers.

Or nothing meaningful will change.

Reality is likely to sit somewhere in the middle.

The software development lifecycle isn't disappearing. It's being redesigned around collaboration between humans and AI agents.

The teams that benefit most won't necessarily be those using the newest models.

They'll be the ones that rethink how work flows through engineering, where human expertise creates the most value and how AI can accelerate the rest.

That's a much bigger shift than writing code faster.

It's a different way of building software altogether.