CTO Predictions for 2026: Special ShipTalk Episode with Nick Durkin

As we look toward 2026, it is becoming clear that AI is not just changing how code is written. It is changing how software delivery itself works. The real shift is happening at the intersection of AI, security, and developer experience, where speed, risk, and responsibility now collide.

Nick Durkin, Field CTO at Harness, recently joined the ShipTalk podcast to share his perspective on where this collision is heading. His predictions are not about flashy tools or futuristic demos. They are about what happens when AI meets the reality of production systems, compliance requirements, and human teams that still have to ship reliably.

At the center of his outlook is a simple idea. AI will not replace software delivery fundamentals. It will expose whether you ever had them in the first place.

From One “Smart” AI to Teams of Specialist Agents

One of the biggest misconceptions about AI today is the belief that a single, all-knowing system will handle everything. Nick expects this idea to fade quickly.

Instead, by 2026, teams will rely on specialized AI agents, each designed to perform a narrow, well-defined role. This mirrors how effective human teams already operate. We do not expect one engineer to be a database expert, a security specialist, and an infrastructure architect all at once. AI will follow the same pattern.

These agents will also validate one another. Rather than trusting a single output, teams will use multiple AIs to cross-check results and establish confidence. This “who watches the watcher” model reduces the need for humans to manually review every change, while still maintaining trust in the system.

As this happens, humans themselves will begin to change how they work. Nick predicts that engineers will increasingly behave like agents too. Focused. Specialized. Spending less time on repetitive toil and more time on creative problem solving, architecture, and business outcomes.

Why Guardrails Will Replace Gates

As AI accelerates development, traditional governance models start to break down. Rigid gates, hard approvals, and one-size-fits-all templates cannot keep up with the pace AI enables.

Nick argues that the future lies in guardrails, not gates.

Many organizations learned the hard way that standardized templates quickly become too restrictive. They cannot account for every valid variation a modern system needs. The result is either constant exceptions or teams working around the process entirely.

By 2026, Nick expects most successful teams to rely on flexible templates combined with policy-driven pipelines. Around 80 to 90 percent of security, compliance, and resilience requirements will be baked in by default. The remaining space allows developers to innovate without breaking the rules.

The goal is not to stop people. The goal is to make it hard to do the wrong thing. Pipelines should automatically enforce policies like secret management or vulnerability thresholds without killing momentum. When something fails, the system should explain why and show the next step forward.

DevSecOps Finally Becomes Real

For years, “shift left” has been more slogan than reality. Nick believes that changes in AI-driven delivery finally force DevSecOps to become real.

When AI can generate changes faster than humans can review them, security cannot sit downstream anymore. It cannot be a separate team delivering reports weeks later. That model simply does not survive.

In teams that are already succeeding, security is fully integrated into the delivery lifecycle. Security teams define policies. Engineers understand the rules. Pipelines enforce them automatically.

Nick compares this to a video game. Vulnerabilities and misconfigurations become obstacles you encounter in real time. You learn, you fix, and you move forward. The feedback loop is immediate, and progress toward production is clear.

This approach removes friction between teams. Security stops being a blocker and becomes part of how work gets done.

Every Engineer Becomes an Engineering Manager

Nick’s most provocative prediction is about the role of the engineer itself.

By 2026, every engineer effectively becomes an engineering manager. Not of people, but of AI agents.

Instead of wrestling with code line by line, developers will manage a collection of agents performing specific tasks. Writing boilerplate, fixing known issues, scanning for vulnerabilities, or updating dependencies. The engineer’s job becomes orchestrating this work and, more importantly, providing context.

That context is where human value concentrates. Business intent. Historical decisions. Tradeoffs. The “why” behind the system. AI does not have this unless we give it to them.

Despite widespread fears, Nick is optimistic about employment. History shows that major technological shifts do not eliminate work. They expand what is possible. Virtualization did not end infrastructure roles. Cloud did not eliminate operations. AI will not eliminate engineers.

It will raise expectations.

The Real Prediction for 2026

AI will not save your software delivery.

It will expose it.

Teams with strong pipelines, clear policies, and shared rules will move faster than ever. Teams without them will argue more, ship riskier, and blame AI for problems that already existed.

AI does not change the fundamentals of delivery.
It removes the excuses.

🎧 Listen to the full ShipTalk podcast

‍