AI is making people more productive and more worried about their jobs, often because of the same thing. This edition we look at what 81,000 workers actually told us about living through that reality, why agent governance is becoming the infrastructure problem no one budgeted for, and why the data stack underneath your AI matters more than the model on top of it.
The trend: A survey of 81,000 Claude users finds that AI is genuinely making people more productive, but the workers seeing the biggest gains are also the most anxious about their long-term job security.
The details: Anthropic's Economic Index survey found that the most common benefit cited was expanded scope rather than just speed: 48% of respondents said AI let them do tasks they previously couldn't, while 40% said it made existing tasks faster. Workers in roles with higher observed AI exposure were more likely to say their job was already being replaced or soon would be, and early-career workers expressed significantly more anxiety than senior professionals. The workers reporting the largest speedups were also the most worried about displacement.
Why it matters: The survey captures something the macro productivity data misses: AI is already changing how individuals work, and the people closest to that change can feel both sides of it at once. The tension between personal productivity gains and job anxiety is going to shape how AI gets adopted, resisted, and governed over the next few years.
The trend: Agentic AI is moving from pilot to enterprise deployment faster than most organizations can govern it, and the gap between adoption and control is turning into a real operational problem.
The details: Gartner predicts that by 2028, the average Fortune 500 company will use over 150,000 AI agents, up from fewer than 15 in 2025. Only 13% of organizations think they have the right governance in place today. Without structure, the result is agent sprawl: ungoverned deployments that expose organizations to misinformation, data oversharing, and regulatory penalties. On the leadership side, Gartner also predicts 45% of CIOs will be leading agent systems outside IT entirely by 2028, making governance a C-suite problem, not just an IT one.
Why it matters: Blocking agent use doesn't work, as employees get around controls and ungoverned AI ends up being the bigger risk. Organizations need to build governance infrastructure now, before the sprawl gets unmanageable.
Stanford's 2026 AI Index found that employment for software developers aged 22 to 25 has fallen nearly 20% since 2022, with customer support roles showing a similar pattern. Senior and mid-career workers in both fields have continued to grow over the same period, suggesting the pressure is concentrated at entry level. A third of organizations surveyed by McKinsey expect AI to shrink their workforce in the coming year, particularly in software engineering and service operations. The productivity gains are real, but they are not being shared evenly, and early-career workers appear to be absorbing most of the risk.
A new MIT Technology Review piece makes a simple argument: the quality of your AI is entirely dependent on the quality of your data, and most enterprise data is still locked in siloed systems with no real governance around it. Databricks SVP Bavesh Patel is blunt about what that produces: "terrible AI." The piece walks through what getting it right actually looks like, from moving data into open formats and building proper cataloging, to governing agent access and measuring business outcomes rather than just adoption rates. The companies pulling ahead are not using better models. They started by getting their data foundations in order, then built from there.
Harvard Business Review found that some patterns of AI use are driving 'brain fry' where users report a "palpable sense of stress" from agent-orchestration systems moving faster than humans can reasonably follow.
%201.svg)
