Meta's CTO Admits the AI Reorg Was Atrocious, Then Offers Bigger Snack Budgets

Welcome back to SavvyMonk, your one-stop for AI and tech news that actually matters. Meta has put an AI answer engine inside Facebook search this week, and at the same time its own chief technology officer is telling staff the mood inside the company is about as low as it has been in twenty years. Those two stories come from the same company in the same stretch, and they pull in opposite directions.

Let's get into it.

104 matches. 48 teams. 39 days of football. Right now, streaming means logging into a cable account you don't have or paying for a subscription you'll cancel in August.

Norton Neo is a free browser with a free built-in VPN. No sign-up, no credit card, no catch. Private by default, backed by Norton security. Anti-fingerprinting and ad blocking run quietly in the background while you watch.

Download in 45 seconds. Watch every match for free.

Fast. Safe. Intelligent. That's Neo.

Try Norton Neo Now

On June 15, Meta rolled out a new search tab called AI Mode inside Facebook, one that answers questions in plain language instead of returning a list of links, drawing its responses from the culture, opinions, and recommendations people post publicly across the company's apps.

The feature pulls from public Groups, Reels, and Marketplace listings, and it runs on Meta AI together with Muse Spark, the first model from the company's Meta Superintelligence Labs unit, which the company introduced back in April. Rather than scrolling through results, a person can ask something conversationally, get a single synthesized answer, and then keep asking follow-up questions in the same thread.

It is the same idea Meta tested last month with Forum, a Reddit-style app built on Facebook Groups that shipped with its own AI Ask tab, and it lands the company squarely in the answer-engine race that Google, Microsoft, and Amazon have been running for more than a year.

The Mood Behind the Launch

The product looked confident, but the mood inside the building did not match it. During an internal session on June 2, Andrew Bosworth, the company's chief technology officer and one of Mark Zuckerberg's longest-serving lieutenants, told staff that morale is probably one of the worst it has ever been across his two decades at the company.

He hedged the claim, saying it was maybe not the absolute lowest point in twenty years but that it was definitely up there, and he reached for the Cambridge Analytica scandal as his point of comparison, which is a striking benchmark given how much reputational damage that episode caused. For one of Zuckerberg's most loyal deputies to say this out loud, even in a closed setting, tells you how widely the unhappiness has spread.

What Set It Off

The discontent has clear roots. In May, Meta cut roughly 8,000 jobs, about ten percent of its global workforce, and reassigned thousands more into a newly formed engineering division called Applied AI, a unit that swept up around 6,500 staff with little say in the move and little clarity about what it meant for their careers.

Workers have described the reassignment as being drafted, and one employee called the division a gulag in conversations with the press, while separate reporting said the unit was on the verge of revolt with engineers calling the work soul-crushing. Adding to the friction, employees have pointed to monitoring software running on company laptops since April that logs their keystrokes and mouse activity to help train AI models, alongside widening pay gaps and the forced moves themselves.

The Money Does Not Match the Mood

What makes the gloom strange is the financial backdrop. Weeks before the layoffs, Meta posted one of the richest quarters in its history, with about 26.8 billion dollars in profit on 56.31 billion dollars in revenue for the first three months of 2026, and sales up thirty-three percent from a year earlier, the company's fastest growth since 2021.

The cuts were framed as a way to fund the company's enormous AI ambitions, which Meta has forecast could reach as much as 145 billion dollars in capital spending for 2026 alone. The pain, then, is not coming from a business in trouble but from a company pouring record profits into an AI bet and reshaping its workforce to match.

What Leadership Is Trying

Bosworth has not ignored the problem. In a memo to staff on June 16, he admitted the company had done an atrocious job explaining the vision behind the Applied AI division and conceded it had undermined the trust people placed in their own expertise, then laid out a set of fixes. He pledged to cap managers at roughly twenty direct reports, to move staff between teams less often during reorganizations, and to offer optional AI coaching, while reassigned workers were told they could now apply for other roles inside the company.

The company has also loosened the purse strings on travel, events, and office perks, increasing budgets for things as small as snacks in an effort to lift the day-to-day mood. The snack line is easy to mock, but it sits inside a broader and more serious attempt to steady a workforce that feels shuffled around without a say.

Why It Matters

Set side by side, the two stories say something uncomfortable about where Meta is right now. The same AI push that is straining morale is also what produced AI Mode, the splashy consumer feature meant to keep people inside Facebook and give the company fresh ways to make money.

And the feature itself carries real risk, because an answer engine that summarizes whatever people post publicly inherits every problem of user-generated content, from stale advice to confident nonsense, the same trap that produced infamous suggestions like gluing cheese onto pizza when rival systems leaned too hard on forum chatter.

Meta is also arriving late to a race its biggest competitors have run for over a year, which raises the stakes on getting both the technology and the people behind it right.

The Bottom Line

Meta is shipping ambitious AI features at the exact moment its own people are most worn down by the AI push driving them. Record profits and a flashy new search tab cannot paper over a workforce that its own CTO says is near a twenty-year low.

Bigger snack budgets and fewer reorgs might help at the margins, but the real test is whether Meta can convince the engineers building its future that they actually want to be there. A company can buy compute, but morale is much harder to buy back.

Category: Change Management

"Act as an executive communications advisor. I am a leader at [Company] rolling out [major change, such as a reorganization or a new AI initiative]. Draft a 300-word message to my team that explains the reasoning behind the change, acknowledges the disruption it causes, and lays out three concrete steps I will take to support people through it. Keep the tone honest and direct, avoid corporate jargon, and end with a clear invitation for feedback."

There is a lesson in watching a company post record numbers and admit record-low morale in the same month. Growth and goodwill are not the same currency, and the second one is far easier to spend than to rebuild. If your own workplace is leaning hard into AI right now, it is worth asking whether the people doing the work feel brought along or simply reassigned. Hit reply, I read every response.

See you tomorrow.

— Vivek

P.S. Know a founder or operator navigating a big AI shift at their own company? Forward this their way. They can subscribe at https://savvymonk.beehiiv.com/