Anthropic recently advised its growth team to hire more product managers. This strategic shift challenges traditional software roles. Claude Code reportedly increased its engineering team’s output by roughly three times its size. The core constraint in software development has changed. It moved away from merely typing code. It now centers on deciding what the code must achieve. This structural change limits the productivity of engineers who only write code. Such engineers may struggle to advance their careers.
For the last decade, software creation followed a clear division of labor. Engineers built the product. Product managers managed the entire user funnel. This traditional setup kept the two roles separate. Generative AI is rapidly collapsing this distinction in five major steps. This swift evolution demands that engineers adopt the mindset of product thinkers.
The Evolution of the Engineer’s Day
Engineers approach problems differently now than they did since 2014. During the Stack Overflow era, engineers kept their problem-solving localized. However, monthly Stack Overflow questions dropped by about 77% since November 2022. This decline aligns with the launch of ChatGPT. It reflects a workflow change, not a site failure. The subsequent browser-tab period saw engineers use AI outside their integrated development environment. They generated prompts in a browser and pasted the answers into their code editor, keeping the work single-threaded.
This localized leverage changed completely with the IDE-native era starting around 2024. Tools like Cursor and Claude Code moved the AI model directly into the editor. This gave the AI access to the entire codebase. This development path dissolved the need to escalate complex issues to a senior engineer. Veteran engineers once believed Bash scripting offered the longest operational lifespan. Yet, some sources claim Claude will often be the first command typed in a fresh terminal by 2026.
The Spec-driven Era Beginning Compressed
The spec-driven era starts in 2025, compressing feature builds drastically. Amazon’s Kiro IDE team reportedly cut feature build times from two weeks down to two days. They used the same specification-driven methods. One AWS engineering team noted a significant achievement. They finished an 18-month rearchitecture in just 76 days with six people. The limiting factor stopped being typing speed. Instead, it became how clearly the team defined what correct functionality looked like. Anthropic then released Claude Code Routines in April 2026. These persistent agents allow engineers to orchestrate complex tasks, bringing back concepts like cron jobs and hooks.
The Decision Bottleneck: Becoming Product Thinkers
engineering output has roughly tripled across many organizations. However, product management capacity remains static. The historical ratio of one product manager to eight engineers is now effectively a 1:20 ratio. Each engineer ships considerably more features daily. Companies deploying agentic workflows in production consistently produce features faster than they produce decisions about those features. This imbalance requires teams to restructure their roles. For example, LinkedIn replaced its associate product manager track with a “Product Builder” program. This program trains generalists across design and engineering. Anthropic’s decision to hire more product managers confirms this industry trend. The new requirement is that engineers must function as product thinkers.
Engineers must now take on tasks historically reserved for product teams. This means moving beyond simply waiting for a Jira ticket. The new role requires engineers to actively engage with the business side of the product. They must start asking questions like these:
Talk to customers to understand how they truly use the product. Read the support queues to identify recurring user pain points. Sit in on sales calls to grasp customer needs firsthand.
An engineer can now gather firsthand product signals in an afternoon. Previously, a product team needed several layers of summarized reports for this task. The capability to generate scoped ideas, rather than just providing estimates, is now a key differentiator.
Why Fundamentals Still Matter as Product Thinkers
Some people suggest that core engineering principles are now obsolete due to AI agents. This view ignores a key technological point. When a memory leak causes a production failure at 3 a.m., no current agent can close that loop end-to-end. Operating systems, network protocols, and concurrency rules still determine who resolves real incidents. These low-level concepts also determine when an agent’s output looks correct on the surface but is quietly wrong beneath the code. The team needs the engineer who can read the code changes and spot those subtle errors. This engineer relies on deep fundamentals, not just prompting skill.
Fundamentals have shifted from being a maintenance skill to a powerful advantage. In 2014, knowing how a TCP retransmit worked sped up a debugging ticket. By 2026, that same knowledge prevents an entire AI-driven release pipeline from shipping a massive error at scale. The potential scope of impact for an engineer who understands underlying systems increased substantially. This makes deep knowledge essential for quality assurance.
Code Generation Speed Now Far Exceeds Human Limits
Code generation speed now far exceeds any engineer’s ability to read every line carefully. Teams that ship quickly and remain stable must treat AI-generated code reviews with the same rigor as writing the code themselves. Data from the 2025 Stack Overflow developer survey showed 84% of developers use AI tools. Yet, 46% stated they did not trust the output. This gap between heavy usage and low trust highlights the importance of review skills. Engineers who produce high volume but review little accumulate technical debt. This debt will appear during the first real incident. The engineer who pays that debt back possesses deep knowledge of the systems involved. This blend of coding speed and systemic knowledge makes the product thinker the most valuable hire today. Related context: AI coverage.
