Target: LinkedIn Posted: https://www.linkedin.com/pulse/why-your-software-engineering-agent-needs-you-think-like-omer-ansari-9vkwc/

Why Your Software Engineering Agent Needs You to Think Like a Platform Engineer

She’s staring at a screen. Same as always. But something’s different this time.

Her AI agent just deployed a three-tier application—frontend, API, database—while she grabbed coffee. It wrote the code. Ran the tests. Fixed the bugs. Shipped it.

And you’re wondering: What exactly is her job now?

Here’s what I’m learning: Engineers aren’t disappearing. We’re evolving into something bigger.

Think about it. Your AI can write a React component faster than you can. But can it distill the real problem from stakeholder chaos? Set up the guardrails? Handle the non-functional requirements? Troubleshoot when the deployment goes sideways at 2 AM?

Not yet. That’s where we come in.

The future engineer does five things:

Problem distillation - Turning messy business requirements into clean technical problems Guardrails & NFR oversight - Security, performance, compliance—the stuff that breaks products Tool integration - Wiring together the ecosystem so agents can actually work Troubleshooting & reframing - When AI gets stuck, we unstick it Environment orchestration - Designing the runtime where agents can think, build, and ship

This post? I’m pulling on that last thread.

Because here’s what I’ve been prototyping: Engineers as environment designers.

Remember when platform engineers served software engineers? Built the CI/CD pipelines? Managed the Kubernetes clusters? Handled the infrastructure so developers could focus on features?

Yeah, that’s flipping.

Now you’re the platform engineer. For your AI agents.

You’re building the scaffolding. The container orchestration. The deployment pipelines. The monitoring dashboards. The rollback strategies. All so your AI can code without breaking production.

You’re not just writing code anymore. You’re designing the platform where code gets written.

Your AI agent needs tooling. Observability. Service meshes. Database connections that don’t timeout. APIs that respond in milliseconds, not minutes. A modular, reproducible environment where it can iterate fast without setting everything on fire.

And when it does set things on fire? You’re the one with the extinguisher.

I’ve been building this with Dagger.io (though Docker and shell scripts work too). An AI agent runs in one container, orchestrating others. It spins up React frontends, Node.js APIs, Postgres databases—whatever the project needs. All programmatically. All reproducible.

The result? My agent can iterate on both code AND infrastructure together.

This isn’t about replacing engineers. It’s about evolving us into product engineers—directing AI agents like conductors directing an orchestra. We set the tempo. We shape the sound. We make sure everyone plays in harmony.

Software engineers and product managers might merge into something new. Call them “product engineers.” People who understand both the technical stack and the business problem. Who can guide AI agents toward solutions that actually ship and actually matter.

Check out the full prototype: https://github.com/knail1/ai-agent-dagger-containers-demo

Which of these five roles feels most critical to you right now? Did I miss any? And what are you building to get better at it?