According to a 2025 report from Ravio, entry-level hiring (especially in engineering roles) has collapsed by more than 73% due to increasing AI capabilities. That means junior developer jobs are disappearing. At the same time, demand for senior engineers keeps climbing because organizations need more people to manage and optimize their complex AI agents.

“You don’t need to be a math genius to realize that’s going to run out at some stage,” said Accenture’s Mirco Hering (Global Offering lead for Developer Experience and Platform Engineering) at PagerDuty on Tour Sydney in 2026. Fewer junior engineers hired today means far fewer senior engineers a few years from now.

But what looks like a talent pipeline problem is also an operations problem. If senior engineers hold the necessary operational knowledge to manage AI systems, what happens when that knowledge disappears in just a few decades? With AI remaining somewhat unpredictable and shipping code faster than anyone can review, the problem presents a major risk to security and resilience.

PagerDuty’s 2025 AI Resilience Survey suggests that organizations are already aware of the problem. Seventy-six percent of companies running one AI agent (79% running multiple) believe AI’s complexity will outpace the people they have to manage it.

The real question is: What can we do about it?

In 2026, prepared IT leaders are using AI to elevate, evolve, and encode engineering skills into a system that uses operational knowledge to automate work for teams that keep getting smaller. Here’s how.

Elevate: Make AI serve engineers, not the other way around

With AI automating production work traditionally assigned to junior developers, the shift in engineering workflows is inevitable. But some companies are realizing they’ve shifted in the wrong direction.

In late April 2026, Meta reassigned as many as half of engineers from core infrastructure, product, and security teams into a new group built to train its AI coding model. More than 4,500 engineers moved off product work and onto data labeling and similar tasks to serve its coding machine. The shift affected morale in the once well-regarded engineering culture and massively reduced staff. Instagram’s design team dropped by 44%, and its developer documentation and support teams by 95%.

This ultimately led to an incident impacting multiple high-profile Instagram users, including Barack Obama’s White House account.

Meta treated AI as its most important engineer and turned human engineers into mere support staff. Innovation and creativity were restricted to what the AI was capable of—which isn’t enough to compete.

Instead, organizations should flip the hierarchy. Let AI absorb the repetitive, low-judgment work, and elevate human engineers to plan, build, manage, and optimize more resilient systems.

For example, Intuit uses AI to generate post-incident reports, so engineers can spend more time on deeper analysis (PagerDuty on Tour 2026, San Francisco). As soon as an incident resolves, AI scans the related conversations and drafts the report automatically. This approach pushes engineering work to a higher level, allowing them to spend more time on complex, creative thinking than they could without the agent.

Evolve: Reconsider how engineers work

To preserve operational expertise, organizations need a culture that shares institutional knowledge instead of siloing it. That starts with a safe engineering culture.

When engineers fear blame, they hide mistakes instead of reporting them. Google’s research and DORA both point to psychological safety as one of the strongest predictors of team effectiveness and lower burnout. Blameless culture is a key part of making this possible. When teams take away the fear of failure, incidents become learning opportunities. As Dave Hogan, Principal Architect for Service Reliability Engineering at Workday, puts it: “We want our engineers to feel comfortable raising risk… before they cause incidents.”

That kind of openness allows for seamless knowledge transfer and apprenticeship. Freed from constant firefighting, senior engineers gain the cognitive freedom to pass down operational context to the next line of developers.

Ultimately, this elevates how junior engineers work. Instead of spending years memorizing syntax, juniors can immediately start learning senior workflows, enabling them to tackle highly sophisticated problems. They can skip the busywork and go straight to system design, auditing, and validation. This creates deeper engagement and ownership of the system.

In fact, this shift is already underway. Engineers are spending less time writing routine code and more time overseeing AI agents, CNN reports. Entry-level demand for AI skills has tripled, according to NACE’s Spring 2026 update. As Accenture’s Mirco Hering puts it, the challenge isn’t writing code quickly anymore. It’s: “How can you control it in a world where your co-workers are agents?”

Build a safe, open mentoring pipeline, and you don’t just protect the expertise you have—you scale it by training the strategic leaders you’ll need tomorrow.

Encode: Build operational knowledge continuously

Making operational knowledge available isn’t enough. As it stands, most of that knowledge gets filed away inside a Jira ticket that no one ever opens again. Instead, your hard-won data and your best engineers’ learnings and expertise should be built into a system that applies (and learns from) it consistently.

PagerDuty’s Scribe and SRE Agents work together to close that gap. Scribe Agent captures new knowledge automatically, ingesting incident meetings and chat threads (Slack and MS teams) to turn conversations into permanent institutional intelligence. SRE Agent then applies that history to investigate, triage, and handle incident response at three levels:

• Smart routing and documentation: For novel problems, routes the issue to the right experts, drafts stakeholder narratives, and logs the resolution to learn from next time
• Assisted triage: Analyzes historical data to surface patterns and suggest proven solutions to the engineer on call
• Autonomous execution: Resolves well-understood, recurring incidents instantly, without human intervention

This way, knowledge doesn’t evaporate when an incident closes or an engineer leaves. Every incident makes the system smarter. And once incident response runs on an automated, self-improving system instead of manual effort, engineers get the real payoff: time back to focus on strategic thinking and creative work.

Resilient operations begin with engineers

AI is automating traditional junior-level work, but this shift shouldn’t leave entry-level talent behind. Real operational resilience is all about building a culture that values human intellect at every stage of the career path.

When AI handles repetitive tasks, teams get the space to focus on growing their staff of engineers into senior systems thinkers—and organizations keep invaluable institutional knowledge in-house.

That builds a self-sustaining culture of growth, where your team’s collective expertise stays with your business instead of walking out the door. Before long, your engineers become skilled architects who know your systems inside and out and provide your operation with a crucial advantage.

PagerDuty helps safeguard that knowledge further. Built-in AI agents continuously learn and surface patterns across your operations, automating more of the toil that used to limit your team’s time. That frees senior engineers to mentor and junior engineers to focus on the strategic, creative work AI can’t replicate.

That future is just around the corner. Download the ebook: From Firefighting to Autonomous Operations to see how your team can get there.