Agentic AI

The Agent-First Enterprise: Training Engineers to Orchestrate, Not Just Implement

September 29, 2025 18 min read
Diverse team of engineers collaborating around holographic AI interface

In this series we have discussed much about how agents are changing the face of software development. But how are we handling the "human angle" of this discussion? This post in my series addresses that topic in more detail.

The role of software developer is undergoing its most significant transformation in decades. With AI adoption among developers surging to 90% in 2025, the profession is shifting from code implementation to system orchestration. This evolution demands a fundamental rethinking of how we train both emerging talent and seasoned professionals. The key differentiator will no longer be the ability to write code line-by-line, but rather the capacity to architect solutions, validate AI outputs, and translate business intent into technical reality.

The Inflection Point We're Facing

We're witnessing a paradigm shift that's caught many technology leaders off-guard. According to McKinsey's 2025 workplace AI report, 46 percent of leaders identify skill gaps in their workforces as a significant barrier to AI adoption. More telling is this statistic: while 50% of HR leaders believe they're excelling in skills training and internal mobility, only 20% of employees agree.

This disconnect reveals a deeper truth—we're still training developers for yesterday's job description.

In my work with enterprise teams across retail, technology, and media sectors, I've observed a consistent pattern: organizations are investing heavily in AI tools but dramatically underinvesting in the human capability required to use them effectively. The World Economic Forum projects that 170 million new jobs will emerge by 2030, but 40% of core skills will fundamentally change for workers. The question isn't whether developers will have jobs—it's whether they'll have the right skills for those jobs.

From Implementation to Orchestration: The Cognitive Shift

The most profound change in software development isn't technological—it's cognitive. Developers are transitioning from being implementers who write individual lines of code to orchestrators who design and validate intelligent systems.

Think of it this way: a traditional developer was a craftsperson, carefully handcrafting each piece of functionality. Today's AI-powered developer is more like a conductor of an orchestra—directing multiple instruments (AI models, code generation tools, APIs) to create harmonious outcomes. The skills required for these roles are fundamentally different.

What Orchestration Really Means

Orchestration in the AI era encompasses several critical capabilities:

Intent Definition: The ability to articulate precisely what needs to be built and why, translating ambiguous business requirements into clear, actionable specifications that AI can execute on.

System Architecture: Understanding how components fit together, recognizing patterns, and designing for scalability, maintainability, and resilience—areas where AI tools still fall dramatically short.

Validation and Quality Assurance: The judgment to evaluate AI-generated code for correctness, security vulnerabilities, edge cases, and technical debt. Google's 2025 DORA report found that while 80% of developers report AI enhances productivity, concerns about code quality persist.

Contextual Problem-Solving: Grasping the broader system, business constraints, and long-term implications that AI—operating without organizational memory or strategic vision—cannot comprehend.

"The role of developers will shift from implementation to orchestration, focusing on problem solving and system design, and ensuring AI tools deliver high-quality outcomes." — Gartner, 2025 Software Engineering Trends

The Junior Developer Challenge: Building Foundations in an AI-First World

The impact of AI on junior developers represents one of our industry's most pressing challenges. LeadDev's 2025 AI Impact Report reveals that 54% of leaders believe AI coding tools will reduce hiring for junior developers, with 18% expecting fewer junior hires in the next 12 months.

This presents a paradox: if AI can handle entry-level tasks, why hire juniors? But as Honeycomb CTO Charity Majors warns, "By not hiring and training up junior engineers, we are cannibalizing our own future."

Critical Skills for Emerging Developers

Junior developers in the AI era need a different skill foundation than their predecessors:

1. Foundational Computer Science Principles

The temptation is real—let AI write the code, and move fast. But research from METR found that experienced developers using AI tools actually took 19% longer to complete tasks in their own repositories. Why? Because understanding fundamentals remains essential.

Junior developers must learn:

These aren't optional "nice-to-haves"—they're the foundation that prevents over-reliance on AI-generated code that junior developers don't understand and can't debug.

2. AI Literacy and Prompt Engineering

Unlike previous generations, today's juniors must develop fluency in working alongside AI from day one. This includes:

3. Critical Evaluation and Validation Skills

A junior developer's most important skill in 2025 isn't writing code—it's knowing whether code is correct. This means:

4. Learning How to Learn

Perhaps most critically, junior developers need meta-skills—the ability to learn continuously as AI tools and best practices evolve. With AI capabilities changing monthly, adaptability becomes more valuable than mastery of any specific tool.

The Mentorship Imperative

Here's the uncomfortable reality: 38% of respondents in the AI Impact Report agreed that "AI tools have reduced the amount of direct mentoring junior engineers receive from senior engineers."

This creates a dangerous gap. Senior developers, busy training AI to generate better code, aren't training junior developers to become better engineers. Organizations must intentionally preserve mentorship even as AI handles more routine tasks. This means:

The Senior Developer Evolution: From Coder to Architect

If junior developers face questions about relevance, senior developers face questions about evolution. The good news: being a senior engineer, as Charity Majors articulates, "is not primarily a function of your ability to write code." It's about understanding, maintaining, explaining, and managing large bodies of software in production over time.

AI amplifies this reality. Bain & Company found that AI delivers 10-15% efficiency gains across developer tasks, but the real productivity comes when organizations redesign entire workflows—leading to 25-30% improvements.

Critical Skills for Seasoned Developers

Senior developers need to deepen capabilities in areas where AI remains weak:

1. Strategic System Design

While AI can generate components, it cannot architect systems. Senior developers must excel at:

2. AI Tool Selection and Integration

Not all AI tools are created equal. Senior developers need judgment about:

3. Technical Leadership and Knowledge Transfer

As code-writing becomes commoditized, technical leadership becomes more valuable:

4. Quality and Security Stewardship

Senior developers bear ultimate responsibility for code quality—whether human or AI-generated:

5. Business-Technical Translation

Perhaps the most underrated skill: bridging business and technology. Senior developers who can articulate technical decisions in business terms—ROI, risk reduction, competitive advantage—become indispensable.

Universal Skills: What Everyone Needs

Regardless of experience level, certain capabilities become universal requirements:

Intent Engineering and Prompt Mastery

"Prompt engineering" has become shorthand for AI literacy, but it's deeper than knowing how to phrase questions. It's about:

Best Practices Understanding

AI tools trained on public code repositories reflect both good and bad practices. Developers must know:

Without this grounding, developers become vulnerable to what I call "persuasive mediocrity"—code that looks reasonable but embodies poor practices.

Ethical AI Usage and Governance

As AI becomes integral to development, ethical considerations emerge:

Practical Implementation: Building Training Programs That Work

Based on implementations across multiple enterprise clients, here's what effective AI-era developer training looks like:

For Junior Developers: The 70-20-10 Framework

70% Hands-On Problem Solving

20% Structured Learning

10% Community Engagement

For Senior Developers: Continuous Evolution

Strategic Upskilling

Leadership Development

Experimental Learning

Organizational Enablers

Training doesn't happen in a vacuum. Organizations must create environments that support learning:

The Skills Gap: What We're Missing

The IBM Institute for Business Value estimates that 40% of the workforce needs to reskill over the next three years. But the AI skills gap isn't just about technical capabilities—it's about mindset.

The biggest gap I observe in enterprise teams is the shift from "I need to know how to do this" to "I need to know how to evaluate whether this is done correctly." This requires developing judgment—something that comes from experience, not from prompting AI tools.

Organizations that bridge this gap are those that:

Looking Forward: The Developer of 2030

What will a successful developer look like in five years? Based on current trajectories:

Technical Capabilities

Cognitive Capabilities

Interpersonal Capabilities

The Bottom Line: Invest in People or Pay Later

Here's the stark reality: according to Deloitte, 94% of executives believe investment in AI will be critical to business success over the next five years, yet 57% of tech leaders say their team's current AI skills are low.

This isn't an AI problem—it's a people development problem.

Organizations that treat AI adoption as merely a tooling decision will struggle. Those that recognize it as a workforce transformation challenge—requiring intentional investment in training, mentorship, and culture—will thrive.

The developers who succeed won't be those who can write the most code. They'll be those who can architect the best solutions, validate AI outputs effectively, bridge business and technology, and continuously adapt as capabilities evolve.

The future belongs to orchestrators, not implementors. The question for your organization is simple: are you training your team for that future?

Taking Action: Where to Start

If you're a technology leader looking to build AI-ready development teams:

  1. Assess Current State: Where are your teams on the orchestration maturity curve?
  2. Define Target Capabilities: What skills will your developers need in 2-3 years?
  3. Build Learning Pathways: Create structured programs for both junior and senior developers
  4. Measure What Matters: Track capability development, not just productivity metrics
  5. Invest in Mentorship: Protect and incentivize knowledge transfer
  6. Experiment Boldly: Give teams space to learn through controlled experimentation
  7. Share Learnings: Build communities of practice around AI-assisted development

The transformation is inevitable. The timeline is compressed. The opportunity—and the risk—are both enormous.

The question isn't whether to act. It's whether you'll act in time.

This article is part of "The Agent-First Enterprise" series exploring how organizations can transform their operations around AI agent capabilities. Connect with me on LinkedIn to continue the conversation about the future of software development.

Matthew Kruczek

Matthew Kruczek

Managing Director at EY

Matthew leads EY's Microsoft domain within Digital Engineering, overseeing enterprise-scale AI and cloud-native software initiatives. A member of Microsoft's Inner Circle and Pluralsight author with 18 courses reaching 17M+ learners.

Share this article: