@FinanceYF5: 1/ Anthropic engineers' per capita quarterly code volume increased 8 times in one year. Fiona Fung, who leads the Claude Code team, said on the Lenny podcast: Coding is no longer the bottleneck. She has been an engineer for 25 years, led Microsoft TypeScript, Faceb…

X AI KOLs Timeline News

Summary

Using Claude Code, Anthropic engineers increased their per capita quarterly code volume by 8 times in one year. Fiona Fung, head of the Claude Code team, said coding is no longer a bottleneck.

1/🚀 Anthropic engineers' per capita quarterly code volume increased 8 times in one year Fiona Fung, who leads the Claude Code team, said on the Lenny podcast: Coding is no longer the bottleneck. She has been an engineer for 25 years, led Microsoft TypeScript, Facebook Marketplace, and now she's back to writing code herself. 👇 https://t.co/xEWgXz86NP
Original Article
View Cached Full Text

Cached at: 06/23/26, 03:58 AM

1/🚀 Anthropic engineers’ average quarterly code output has increased 8-fold in a year

Fiona Fung, who leads the Claude Code team, said on the Lenny podcast: Coding is no longer the bottleneck.

She has been an engineer for 25 years, led Microsoft’s TypeScript and Facebook Marketplace, and is now personally back to writing code. 👇 https://t.co/xEWgXz86NP

2/ Once the bottleneck disappears, it’s all about ambition

Fiona’s exact words: AI raises the ceiling on what anyone can do—in theory, everything is possible.

An engineer with no mobile experience used Claude to add an App-side feature directly.

3/ She uses Claude to review the entire team every month

She connected all code repositories and Slack channels, live-streamed the screen with her team to review: what shipped, market feedback, and whether there are quality hotspots.

The tool has evolved from “generating PRs” to “having conversations with people.”

4/ Focus shifted from reviewing code to defining “what good looks like”

Throughput is too high for humans to review. Her approach: define standards as a spec and check it into the repo, so Claude can automatically validate against the framework.

She says this is the evolution of test-driven development: before, writing tests felt like eating broccoli first; now, AI handles it automatically.

5/ Quality ensured through monitoring, not more review

Each team defines what counts as a critical error (crash), minor (UI flicker), etc.

The team even built a “swear word dashboard” to track how often users get angry enough to curse.

6/ Hiring only two types of people

One is “dreamers” with product sense—able to go from idea to polished launch; the other is deep system experts who can tackle hard problems.

She also requires every manager to first work as an frontline engineer—get into the code before managing people.

7/ The only differentiator for those who adapt well

Fiona says people who do well have a growth mindset and are proactive; those who struggle are mostly driven by fear.

Her advice is simple: get close to the thing that scares you—ask yourself, what can I control?

The engineering leader behind Claude Code, Fiona Fung, with 25 years of engineering experience. How does she view product teams in the AI era?

The Lenny podcast dives deep into several topics: how to plan after code output increases 8x with AI, what the “loneliness of context switching” is for engineers, and which roles AI will disrupt next.

Video link:

That’s everything.

If you like this topic:

  1. Follow me (@FinanceYF5)
  2. Like + retweet the first post below

The hotter the data center, the more power it saves?

NVIDIA Rubin server coolant can reach 45°C—hotter than a hot tub.

Traditional data centers rely on cold air, with cooling accounting for up to 40% of total electricity consumption. The new logic: liquid directly touches the chip to carry away heat, while outdoor dry coolers provide natural heat dissipation—zero fans, water consumption drops to near zero.

A 50MW data center can save over $4 million per year on cooling alone.

Data centers no longer need to feel like freezers.

Similar Articles

@FinanceYF5: Anthropic's Boris Cherny points out that programming is moving towards a higher level of abstraction. A year ago, developers' workflow was to first write code in an IDE, then prompt multiple Claude instances in parallel. Now, the workflow has evolved to writing loops to call Claude, and letting it make autonomous decisions...

X AI KOLs Following

Anthropic's Boris Cherny points out that programming is moving towards a higher level of abstraction, with workflows shifting from manually writing code to letting Claude make autonomous decisions, and predicts that the next paradigm shift will arrive this year.

@FinanceYF5: Source:

X AI KOLs Following

Anthropic is reportedly using 1,000 freelance software engineers at $280 per task to train Claude Code, having them write prompts, compare code, and test follow-ups to teach the AI how real developers work.

@ZaynHao: Recorded a video version, Fiona talked about how they work at Anthropic. - The bottleneck has shifted; coding is no longer the most expensive part, processes are unnecessary - A six-month roadmap might be too long - Team composition and role ambiguity - Keep the organization flat, deep dog…

X AI KOLs Timeline

Anthropic employee Fiona shares the company's way of working: code is no longer the bottleneck, processes are simplified, teams are flat, and product managers first act as individual contributors. The article also compares the software delivery approach of early Microsoft, which relied on CDs.