I Watched AI Agents Eat Software Engineering This Week — Here's What Comes Next

The Warning Nobody in Tech Wants to Hear

This week, Anthropic's top engineer Boris Cherny said something that stopped me mid-coffee: AI agents will transform every computer-based job in America, and it's going to be "painful." He wasn't talking about some distant future. He was talking about now.

I've spent 20+ years in tech — shipping quality engineering systems at Amazon, Ring, Tinder, and now Motorola Solutions. I've survived the shift from manual QA to automation, from scripted tests to AI-generated ones. But what I'm watching in February 2026 feels categorically different. This isn't a new tool. It's a restructuring of what work means.

Let me break down the three biggest signals I tracked this week — and what I think every tech professional needs to do right now.

Signal #1: The "Centaur Phase" Is Here

Axios dropped a piece this week calling the current moment AI's "centaur phase" — a chess metaphor where a human paired with a computer beats any standalone player. The insight: an engineer working with an AI agent is now the most powerful unit in the knowledge economy.

Nearly 50% of all AI agent activity right now is concentrated in software engineering. That's not because engineers are lazy — it's because coding is the domain most amenable to autonomous agents. Code has clear inputs, outputs, and test criteria. It's automatable by design.

As someone who has built Playwright automation agents, self-healing test frameworks, and LLM-assisted QA pipelines, I feel this in my day-to-day. A year ago, I was directing AI tools step-by-step. Today, I'm reviewing what autonomous agents already did. That cognitive shift — from doing to reviewing — is massive. And most people aren't ready for it.

Signal #2: The "Freelance Agentic" Economy Is Being Born

USA Today published research this week about a new economic class forming in 2026: the Freelance Agentic. These are solo operators using a suite of AI agents to do the work of a 20-person agency.

Think about that for a moment. One person with the right stack can now match the output of an entire department. I've experienced a version of this myself — using AI-driven automation at Motorola Solutions to get 40% efficiency gains that would have required multiple additional hires just two years ago.

The implication for the job market is uncomfortable but honest: it's not that jobs are disappearing overnight. It's that the leverage ratio is exploding. Companies can do more with fewer people. That changes hiring, compensation, and what it means to be "senior" in your field.

For Suneet Malhotra QA automation practitioners — and for tech professionals broadly — the question is no longer "will AI affect my job?" It's "am I building the skills to be the orchestrator, not the automated?"

Signal #3: Anthropic Is Going All-In on Enterprise Agents

Also this week: Anthropic launched a major push for enterprise agents with plugins spanning finance, engineering, and design. This is Claude moving from chat interface to agentic infrastructure — embedded directly into the workflows where knowledge work happens.

This matters because it signals where the market is going. OpenAI, Anthropic, and Google are no longer competing on chatbot quality. They're competing on who owns the enterprise automation layer. The winner of that race will reshape how companies run.

For quality engineers, this is both threat and opportunity. As AI agents take over execution — running tests, triaging bugs, generating reports — the value of QA shifts upstream. The people who will thrive are those who can define what quality means, govern how agents behave, and catch what automated systems miss.

What I'm Doing Right Now (And What You Should Too)

I'm not panicking. I'm adapting — fast. Here's my current playbook:

1. Get fluent in agent orchestration. Not just using AI tools, but understanding how to chain agents, evaluate their outputs, and set guardrails. This is the engineering skill of 2026.

2. Double down on judgment. The scarce resource in an agentic world isn't execution — it's knowing what good looks like. Deep domain expertise in QA, architecture, or system design becomes more valuable, not less.

3. Own the quality governance layer. As test execution gets automated, someone has to own test strategy, coverage decisions, and risk assessment. That's where Suneet Malhotra AI testing leadership lives.

4. Build in public. The engineers and QA leaders who are sharing what they're learning — on LinkedIn, GitHub, through blog posts like this one — are building the reputations that will matter in a compressed, agent-augmented market.

The Bottom Line

The centaur phase is real. AI agents are eating software engineering workflows in real time. Anthropic's own data says 50% of agent activity is already in our industry.

This isn't a reason to fear the future — but it is a reason to move faster than you're comfortable with. The professionals who thrive in 2026 and beyond won't be the ones who resisted the shift. They'll be the ones who learned to direct the machines better than anyone else.

I'm building that skill set every day. Are you?

---

What's your take on the centaur phase? Drop a comment or connect with me on LinkedIn — I read every message.