Claude Code triples productivity: now the bottleneck is PMs
Generative AI has shifted the bottleneck in software development: it's no longer writing code, but deciding what to build. Companies are starting to hire more product managers.
June 29, 2026 · 4 min read
TL;DR: Claude Code has tripled the productivity of Anthropic's engineers, moving the bottleneck from development to product design. Companies now need more product managers to think about what to build, not just how to build it.
The quiet shift at Anthropic
Anthropic recently asked its growth team to hire more product managers, not fewer. The reason, reported by VentureBeat, is that Claude Code had transformed its engineering team into a force producing roughly three times what its actual size would suggest. The bottleneck moved from the integrated development environment (IDE) to the people who decide what to build.
This detail goes unnoticed in the noise of AI productivity claims. Yet it represents the structural shift the rest of the industry is starting to experience. The bottleneck in software is no longer typing. It's deciding what to type. And engineers who treat that as someone else's problem are about to stall.
A brief history of how the engineer's day compressed
The Stack Overflow era (2014 to late 2022)
The way engineers thought resided in a single place. But new monthly questions on Stack Overflow have dropped roughly 77% since November 2022, right when ChatGPT launched. The drop is not a referendum on the site, but on the workflow it represented.
The browser tab era (late 2022 to 2024)
The first generation of ChatGPT was outside the IDE. Engineers ran the same cycle as always, but with a faster oracle: write a prompt in the browser, paste the answer into VS Code, repeat. The work was still single-threaded and engineer-driven. The leverage was real but local.
The IDE-native era (2024 to 2025)
Cursor and Claude Code moved the model inside the editor and gave it access to the full repository. The scaling path to senior engineer largely dissolved. For years, the prevailing wisdom among veteran engineers was that Bash had the longest shelf life of any tool in the stack. By 2026, for a significant portion of active developers, the first command typed in a new terminal is claude.
The spec-driven era (2025 to 2026)
Larger context windows turned single-session work into something that previously required tickets, design documents, and sprints. Amazon's Kiro IDE team compressed features from two weeks to two days using the same spec-based workflow. An AWS engineering team described an 18-month rearchitecture, originally planned for 30 engineers, completed by 6 people in 76 days. The bottleneck stopped being how long it takes to write the code and became how clearly the team can describe what correct looks like.
The routines era (2026)
In April, Anthropic launched Claude Code Routines: persistent scheduled agents that run on a cron schedule, on a webhook, or overnight while the laptop is closed. The engineer's job is now partly orchestration: launch a swarm before bed, review a pile of pull requests in the morning. Third-party wrappers like OpenClaw, briefly suspended by Anthropic in April before being partially reinstated, signaled the same from the open-source side.
The bottleneck moved; most teams haven't
Engineering has roughly tripled. Product management hasn't budged. The traditional ratio of one PM per 5-7 engineers no longer makes sense when each engineer produces like three. Companies are starting to feel the friction: great code for the wrong product. Demand for PMs with deep product thinking is rising, as evidenced by Anthropic's decision.
What this means for engineers and companies
For individual engineers, the most valuable skill is no longer pure technical prowess, but the ability to define problems, prioritize, and translate business needs into clear specifications. Those who refuse to develop these skills risk being replaced by AI agents executing specs. For companies, AI investment must be accompanied by restructuring product teams, hiring more PMs, and training engineers in product thinking.
Long-term implications
This shift could lead to greater specialization: engineers focusing on agent orchestration and PMs focusing on strategic vision. It could also accelerate the emergence of startups offering product management as a service, given that demand outstrips supply. The open question is whether the market will train enough quality PMs in time, or whether we'll see an even sharper bottleneck in product definition.
The bottleneck in software is no longer typing. It's deciding what to type.