Meta’s AI-for-work lead exits amid workplace transformation push

Emily Dalton Smith’s exit is a reminder that the hardest part of AI at work is not the model, but the operating system around it. Meta tapped her to lead product work for its internal AI transformation, then saw her leave after about two months, just as the company was trying to make AI the center of how employees build, review, and ship products.

Smith was responsible for the interfaces, platform components, memory systems, automations, and shared product experiences that would make AI useful across Meta, including Metamate, the company’s internal enterprise assistant. She had previously led product for Threads and earlier served as vice president of product management at Meta. The departure lands in the middle of Meta’s broader restructuring push, which has already stirred employee backlash over layoffs, transfers, and the company’s approach to AI-driven workplace change.

For monday.com, the signal is bigger than one executive move. It shows how quickly an AI-workplace strategy can run into friction when product ownership, user trust, and internal adoption are not fully aligned. If AI is meant to become the starting point for research, prototyping, sales presentations, or routine work, product teams need a roadmap that is durable enough for engineers to build against and clear enough for sales to explain to customers. Buyers are no longer just asking whether an AI feature can generate output. They are asking who controls permissions, how memory works, and whether the tool is a productivity layer or a replacement layer.

That tension matters inside monday.com, where the company has been making its own pivot into AI. On May 6, 2026, monday.com said it had become an AI Work Platform and said people and agents would work together. Its investor relations materials say more than 250,000 customers worldwide use the platform. In the first quarter of 2026, monday.com reported revenue of $351.3 million, up 24% from a year earlier, and highlighted the launch of its AI Work Platform with Native Agents. The company also said monday vibe was the fastest product in its history to top $1 million in annual recurring revenue, and that new products accounted for more than 10% of total ARR in the third quarter of 2025.

Meta, meanwhile, has been pushing AI deeper into its own operations. In March 2026, it said AI was now part of its risk review process, and it has also outlined plans to develop and deploy four new generations of MTIA chips over the next two years while pursuing a long-term agreement with AMD for up to 6GW of Instinct GPUs. Smith’s departure shows how much execution risk sits beneath those ambitions. For companies building workflow software, the lesson is plain: AI only scales when the product, the process, and the people behind it stay aligned.