Mark Zuckerberg wants Meta to become an “AI-centric” company. An internal document reveals one way the CEO plans to achieve that goal. The company has set specific metrics for how much work certain employees should be doing with AI tools, including when writing code.
Mark Zuckerberg wants Meta to become an “AI-centric” company. An internal document reveals one way the CEO plans to achieve that goal. The company has set specific metrics for how much work certain employees should be doing with AI tools, including when writing code.
According to an internal document seen by Business Insider, Meta's creative tools development department (Creation Org) has set a goal for the first half of 2026. It expects 65% of engineers to write more than 75% of their committed code using AI. "Confirmed code" is code that has been saved and entered into a project's accounting system.
The Scalable Machine Learning unit, which specializes in AI models and infrastructure, aimed to achieve 50–80% of code generated using AI by February 2026. Alongside this goal, a senior development manager commented: “We don’t track this with metrics.”
The document also lists several company-wide goals for the fourth quarter of 2025 for the Central Products division, a horizontal structure that covers Messenger, WhatsApp, Facebook and other key products. One goal is to get 80% of mid-level and senior-level engineers using AI tools like DevMate, Metamate and Google Gemini. It notes that the focus is on “tool adoption” rather than the percentage of code written in AI.
It is stated that 55% of code changes made by software engineers in all central product development departments should be “made with the help of agents.”
Corporate-wide goals for Q4 2025 (Core Products):
55% of changes to software engineers' code must be made using agents.
80% of mid- and senior-level engineers should implement common AI tools.
Scalable ML Team Goal (February 2026):
from 50% to 80% of code created with the support of AI.
Creation Org (Development Division) 1st half of 2026:
65% of engineers are expected to write more than 75% of their committed code using AI.
Водночас зазначається, що основна увага приділяється саме «впровадженню інструментів», а не відсотку коду, написаного ШІ. https://geometrydashwave.com