Meta Cut the Managers. The Work Stayed.

Meta spent much of this year betting that AI meant it needed fewer managers. In its Applied AI organisation, some managers had as many as 50 people reporting to them. Former managers lost their reports and went back to individual contributor roles. The logic was that if AI does more of the work, teams need less coordination.

Code changes rose 220% but features reaching users rose only 36%.
At the same time, major incidents rose 40%, and time spent firefighting them rose 70%.

So this week Meta started asking some of those individual contributors to become managers again.

A team can only move as fast as the organisation around it. AI made producing code cheap. It did nothing for the decisions, reviews, context and coordination that turn code into something a customer uses. Meta removed the people doing that work and left the need for it in place. So the work queued up, and some of it didn't get done. That's where the gap between 220% and 36% went, and where the incidents came from.

I'm still bullish that in the long term teams will be more autonomous, but it only works when you remove the work managers were absorbing. Do that first and you'll need fewer managers as a result. Cut the managers first and the coordination still has to happen somewhere. At Meta, a lot of it turned up as firefighting.

AI really is helping teams build faster, but the bottlenecks around approvals and coordination still exist. That's why the goal needs to be zero blocking dependencies from idea to satisfied customer.

This Week’s Updates

Enabling the Team

AI in Product teams: In 2026, the growing impact on collaboration by Anna Lefour
AI adoption is moving from individual tools to team-wide strategy, reshaping roles, collaboration and governance across product teams.

AI Transformation Requires Redesigning Work, Not Cutting Roles by Faisal Hoque, Tom Davenport and Paul Scade
AI creates more value when organisations redesign roles, workflows and decision rights instead of treating headcount reduction as the strategy.

Product Direction

Your Product Didn't Get Worse by Waldek Mastykarz
Agents can shift customer expectations before satisfaction or churn moves, so teams need to research how the work is changing before adding an AI layer.

AI Evals: A Hands-On Guide for Product Teams by Teresa Torres
Turn subjective AI quality into a repeatable feedback loop using error analysis, golden datasets, code assertions, LLM judges and customer feedback.

Continuous Research

Don’t Outsource the Learning: Why Human-Led Research Still Matters in the Age of AI by Maria Rosala
AI can accelerate research outputs, but teams still need to observe, question and interpret evidence together to build shared understanding.

When Everyone Builds, How Do You Keep Customers at the Center? by the Dscout Team
Leaders from Expedia, Cisco and AWS show how research can govern distributed insight generation without losing customer context.

Continuous Design

Limit the Number of Details by Geoff Teehan
Every feature and exception consumes attention. Reducing detail creates the space teams need for stronger product craft.

Is the Future of Software Certain Once AI Is Everywhere? from Arin Bhowmick
As software becomes more autonomous, designers need to define the judgement, boundaries and handoffs that determine when systems act, ask for help and return control.

Continuous Development

Responsible AI Adoption Needs Developer Workflow Design by Gleb Tsipursky
Responsible adoption requires workflow-level guardrails, named owners and psychological safety, not policies that teams work around.

The End of Programming by Paul Dix
AI shifts engineering from reviewing every line to designing architectures, verification systems and feedback loops that supervise outcomes.

This Week's Poll

Last week’s poll: Keeping evidence and interactions genuinely human led with 41.7%. Choosing what deserves attention and setting clear ownership and standards each received 25%, while defining what complete and valuable means received 8.3%.

Seen an interesting article online? Share it with us, and we might feature it in our next issue!
Click here to share an article

From AI Pilots to Organisational Change

What It Takes To Turn AI Into Real Product Impact.

Most teams have experimented with AI by now. The harder question is what it takes to turn those experiments into lasting product impact, and what needs to change inside the organisation to support that shift.

At UXDX San Francisco, Maz Brumand from ŌURA will explore how teams can move AI beyond isolated pilots and into real business value, while Jod Kaftan from NTT DATA will look at how new interfaces are reshaping the way organisations work. Explore the agenda to see both sessions and how leaders are preparing teams for the next phase of AI adoption.

Explore the San Francisco agenda for 5-6 November, and use NEWSLETTERSF26 for 10% off tickets.

FREE COMMUNITY EVENTS

Video Of The Week

Designing for the Agent Experience: When UX and DX Collide in the Age of AI

In this UXDX talk, Dana Lawson from Netlify shares how teams can approach agent-driven building with trust and safety as the foundation. As AI gives more people the ability to create, ship and change software, product and engineering teams need systems that support speed without hiding critical risk.

Dana walks through a practical blueprint for safer autonomy, covering structured errors, event-driven signals, legible architecture, sandbox execution, human approval flows and fast rollback. This is not theory. It is what happens when agents have real access to real platform primitives.

Watch the full session to learn how teams can keep builders moving while designing the guardrails needed for AI-powered delivery at scale.a Lawson, CTO at Netlify, explores what happens when products are designed for AI agents as well as people.

She shows how clearer architecture, stronger guardrails and intent-driven interfaces improve the experience for human builders too.

Want to go into how careers and leadership are shifting as AI compresses the ladder? My ebook Managing Your Career In The Age Of AI explores how to build judgement, relationships and influence in a world that keeps trying to automate the surface of the work.