Anthropic Hires More Product Managers After Claude Code Triples Engineer Output

In recent months, Anthropic has made a decision that may seem counterintuitive in an era dominated by generative AI. Instead of reducing staff, the company told its growth team to hire more product managers, not fewer. The reason, as reported by industry sources, is that its tool Claude Code has quietly turned every engineer into a team that ships at roughly three times its actual headcount, shifting the bottleneck from the integrated development environment (IDE) to the people deciding what to build.

The bottleneck moves from writing code to product decisions

This detail is easy to miss amid the noise of AI productivity claims, but it represents the structural shift the entire industry is now undergoing. The bottleneck in software development is no longer typing code; it is deciding what to type. Engineers who treat this decision as someone else's problem are about to plateau. For most of the past decade, that decision sat with someone else: the product manager owned the funnel, the engineer owned the build. Both sides treated this division as physics, but today the ratio has changed dramatically.

The compression of the engineer's day can be broken into five phases. The first, the Stack Overflow era (2014 to late 2022), saw engineers solve problems through that forum, but new monthly questions on Stack Overflow have dropped roughly 77% since November 2022, coinciding with ChatGPT's launch. The second phase, the browser-tab era (late 2022 to 2024), brought ChatGPT into the IDE: engineers wrote prompts and pasted answers into VS Code, a still single-threaded loop. The third, the IDE-native era (2024 to 2025), saw Cursor and Claude Code move the model inside the editor, largely dissolving the senior-engineer escalation path. The fourth, the spec-driven era (2025 to 2026), compressed feature builds from two weeks to two days: Amazon's Kiro team reported an 18-month rearchitecture, originally scoped for 30 engineers, completed by 6 people in 76 days. The fifth, the routines era (2026), introduced Claude Code Routines: scheduled, persistent agents that run overnight, on a webhook, or while the laptop is closed.

Tripled productivity but product management has not budged

The result is that engineering productivity has roughly tripled, while product management has not increased. The traditional 1:8 ratio of PMs to engineers, already strained, now plays out closer to an effective 1:20. LinkedIn replaced its associate product manager track with a "Product Builder" program that trains generalists across product, design, and engineering. Anthropic is hiring more PMs, not fewer. The pattern is consistent across companies that have deployed agentic workflows in production: the system produces built features faster than it produces decisions about what should be built.

First principles matter more, not less, in the agent era

There is a common misconception that fundamentals are obsolete in the agent era. The opposite is true. When a memory leak takes down production at 3 a.m. due to a subtle bug pushed four years ago, no current agent closes that loop end-to-end. Operating systems, networks, concurrency, and query plans still determine who can resolve a real incident. The engineer who can read a diff and catch the error is built on fundamentals, not on prompting skill. Knowing how TCP retransmit works in 2026 can prevent an entire agent-driven release pipeline from shipping a regression at scale.

Review is the new writing

Engineers generate code at a rate that exceeds their ability to read it carefully. According to the 2025 Stack Overflow developer survey, 84% of developers use AI tools, but 46% say they do not trust the output, up sharply from 31% the year before. This gap, heavy use paired with low trust, is exactly where review skills now matter most. Coders who push lots and review little are accumulating a debt that will come due during the first real incident. The engineer who can pay it back is the one who paired volume with deep first-principles knowledge of the systems involved.

The new differentiator is the product funnel

Both of those are necessary; neither is sufficient. The engineer who matters in 2026 is the one who has stopped waiting for the funnel to arrive in the form of a Jira ticket. That means talking to customers, watching how they actually use the product, reading the support queue, sitting in on sales calls. Generating ideas, not just estimates. Working backwards from the customer, as Amazon's "press release first" discipline teaches. The honest answer to "Do you have capacity for this idea?" used to be no. With routines, hooks, and a cooperative agent stack, the honest answer is closer to "What is the idea worth?" And that is a much harder conversation to have without a real point of view on the customer.

The five-phase history above is not a history of tools. It is a history of which part of the job a human still has to do. The part that remains human, and will remain so for the foreseeable future, has moved up the funnel: from typing to reviewing to deciding, all the way to choosing the customer to serve and the problem to solve. Engineers who internalize this will spend the next decade doing the most interesting work software has ever produced. Those who wait for a ticket will watch it get written by the agent next to them.

For more on AI API restrictions, read OpenAI tightens API restrictions in Europe. To compare ERP tools for product decisions, see SAP NetSuite vs Odoo. For technical background, read on Wikipedia.

Source: https://venturebeat.com/infrastructure/claude-code-turned-every-engineer-into-three-now-companies-need-more-product-thinkers