The Post-Coding Era:

Who Builds Software Now?

TL;DR

Your software development team is too large, too specialized, and structured for an era that is ending. The post-coding era does not eliminate software roles — it redefines them around domain expertise, architectural judgment, and new tools and processes. Not just developer roles. The entire product development team — project management, QA, documentation, interface design — is being restructured. Organizations treating this transformation as staffing optimization will underinvest in the structural change required. Organizations treating this time as a workforce crisis will overreact and lose institutional knowledge they cannot rebuild. This is a fundamental redefinition of who belongs on product teams and which human skills create value when code production is automated. Our direct experience — including a client engagement where a 60% headcount reduction maintained full delivery velocity — shows the people who thrive in this environment are systems thinkers with deep domain knowledge, not narrow technical specialists. Leaders who avoid these workforce decisions are not avoiding disruption. They are choosing to be disrupted on someone else’s timeline.

Introduction

For fifty years, the software industry built its workforce model on a simple assumption: building software requires large teams of trained specialists performing labor-intensive, cognitively demanding work. Compensation structures, career ladders, hiring pipelines, and university curricula all reinforce this model. A junior developer writes code, gets mentored, grows into architecture, and eventually leads teams of people doing the same thing. The entire talent ecosystem — from computer science departments to coding bootcamps to staff augmentation firms — exists to feed this machine.

That machine is stalling — and nearing obsolescence. As we argued in “The Post-Coding Era”, the mechanical act of translating known solutions into working software is being automated. The response cannot be incremental adaptation. Human software teams operated as assembly-line operations for decades — large, specialized, sequential. The post-coding era replaces them with small, light teams that move fast and innovate. This paper examines what that means for the people on those teams — not in theory, but from direct experience restructuring product development organizations. The people implications are the hardest part of this transition, and leaders who hedge on them will find the market does not hedge back.

From enhancement to displacement: how quickly the ground shifted

CODE Exitos began its AI-assisted development research in 2024 under the internal code name “X-24.” The X stood for “exoskeleton” — our operating assumption, shared by most of the industry, was AI would enhance developers. We continued this framing into X-25. By the second half of 2025, the capabilities had moved beyond enhancement. Our experience converged with what industry leaders were reporting: agentic coding tools were producing work as good as — or better than — most human developers. This was not a gradual realization. The ground had shifted under an assumption we had held for over a year.

This matters for how leaders interpret the research landscape. Studies measuring AI coding tool productivity from 2024 — including widely cited work from DORA, Faros AI, and METR — measured a generation of tools now several capability doublings behind. Their findings about organizational patterns remain valid: the DORA finding that tool adoption without structural change worsens outcomes describes a leadership failure, not a tool limitation. But any study measuring what AI tools can do from 2024 or early 2025 describes a world that no longer exists. For current-state capability evidence, practitioners’ direct experience is the most reliable source available.

What a 60% reduction actually looks like from the inside

One client’s experience makes the transition concrete. The company is a multi-billion-dollar publicly traded manufacturer. Revenue reversals forced the CEO to demand immediate savings across every function. In technology, adversity met preparedness — the EVP of Technology had already moved past AI-assisted coding into agentic development. This was not an experiment interrupted by crisis. It was a capability crisis accelerated.

The first reduction cut approximately 35% of a product development team of just under 100 people — a team spanning every discipline in the product development lifecycle, not just coding. Some cuts addressed underperformance. Many impacted long-tenured employees who were offered alternative roles where possible. Three months later, the CEO mandated a second cut. The team dropped to roughly 40% of its original size.

After three months in this new model, delivery output has stabilized and meets the demands of the business. The team operates in spec-driven development: clearly defined requirements, produced by people with deep domain knowledge, fed into an evolving agentic harness that builds — not just codes — across the full product development lifecycle.

The most effective contributor on the restructured team is a woman with 35 years of operating experience in the business and no coding background. She defines, with precision and from first-hand experience, exactly what the business needs. With the help of LLM tools, her requirements are clarified and rationalized. Unencumbered by the need to translate her business knowledge through human developers, she drives weekly change and improvement in internal systems. Fewer connections in the telephone game means higher-fidelity output. This is not an anomaly. It is the pattern the post-coding era produces: domain expertise and architectural judgment outweigh coding skill.

It is the entire team being rethought — not just developers

The post-coding transformation reaches well beyond developers. Project management, QA, documentation, interface design, and DevOps are all being restructured or eliminated. Our experience is direct: hybrid teams leveraging agentic AI need fewer ancillary functions than ever before. Project management overhead shrinks when delivery cycles compress from weeks to days. Manual QA disappears when AI generates and runs test suites continuously. Documentation becomes a byproduct of the development process, not a separate function.

The argument reduces to five core claims:

• Entire role categories are being eliminated, not just trimmed. When a company’s project management work can be automated, it is not the individual PMs who are affected — the function itself contracts. The same applies to manual QA, task-level coordination, and routine documentation.
• The cascade effects extend far beyond the technical team. Eliminate a layer of project managers and the manager of those PMs has no team. Shrink the development organization by half and the HR business partner supporting it, the recruiter sourcing for it, and the finance analyst budgeting for it all serve a smaller footprint. The implications are wide-ranging and the transformation is just beginning.
• Systems thinking and domain expertise now outweigh coding skill as a hiring criterion. The most valuable contributors to restructured teams understand what the software needs to do and why — not how to write the code implementing it. New hires gaining traction are systems thinkers, entrepreneurs, and innovators who demonstrate mental agility and judgment, not fluency in any specific language or framework.
• The junior-to-senior pipeline is breaking with no replacement in sight. If entry-level coding roles disappear, the apprenticeship model producing senior engineers disappears with them. Gartner estimates 80% of the software engineering workforce will need to upskill by 2027 — but “upskilling” implies the same people doing enhanced versions of the same work. What we are seeing is a redefinition of the work itself.
• Technology leaders — not HR — must own these decisions. A growing trend positions Human Resources as the decision-maker for AI-era team restructuring. This is a mistake. HR can facilitate transitions, manage compliance, and support affected employees. But the decisions about which roles remain, what new team compositions look like, and how development methodology must change require technical judgment HR does not possess. Every major restructuring producing real results in 2024–2026 — Shopify, Klarna, Salesforce, Block — was driven by the CEO or technology leadership, with HR executing the operational mechanics. Technology leaders who abdicate these decisions to HR are not delegating. They are avoiding the hardest part of their job.

What leaders owe their teams — and themselves

Assuming today’s capabilities as a baseline — and they will only advance — the workforce question is not whether to act but how aggressively. Over more than thirty years of working with software product teams in organizations ranging from venture-funded startups to the Global 500, our assessment is direct: 30% or more of current development headcount can be reduced today. For organizations aggressive enough to restructure fully, 50% or more is achievable. We are not in a market of 10–15% trims.

Reduce headcount. Reorganize teams. Reset which human contributions add value to the team and to the business. The window for deliberate action is open now. It will not stay open indefinitely.

Conclusion

The post-coding era demands a different kind of team. Smaller. More experienced. Anchored in domain knowledge and systems thinking rather than narrow technical specialization. The entire product development organization is in scope — not just developers, but project managers, QA engineers, documentation specialists, and the management layers supporting them.

The human cost is real. Job functions — and entire roles — will be eliminated. No amount of euphemistic language about reskilling makes this less disruptive for the individuals affected. The honest conversation — the one building trust rather than eroding it — acknowledges this directly and then focuses on what comes next: which people on your team can adapt, what do your economic circumstances allow, and what is your organization’s appetite for risk and change? These factors will test the courage and decisiveness of technical leadership and demand action. Doing nothing is no longer tenable.