AI promised less work. January delivered more.
A month of reality checks on productivity, agents, and what happens when companies stop chasing AI moonshots.
AI was supposed to save everyone time. Three posts on where the saved hours actually went.
The moonshot narrative dominated January. The real gains came from thinking smaller.
Half the workforce doesn't feel safe enough to say AI isn't working. The bottleneck was never the technology.
Spotlight
The gap between what AI promised and what showed up. And the part nobody is managing.
The AI productivity paradox
Executives say AI saves them a full workday every week. Their employees say almost nothing. Both are telling the truth, and that's the problem.
Managing AI agents
Companies are deploying AI agents with no one managing them. A framework for treating them like any other hire.
What we read this month
Executives say AI saves them eight hours a week. Two-thirds of employees say less than two. Now there are numbers.
PwC surveyed 4,454 CEOs. More than half report zero gains. The chairman's diagnosis: they skipped the basics.
OpenAI's CFO calls it a 'capability overhang.' AI can already do far more than most companies capture. The bottleneck isn't the models.
A research paper argues transformers are structurally incapable of reliable agentic behavior at scale. The industry disagrees.
Smaller models, real workflows, systems designed for humans instead of benchmarks. The pivot from bigger to usable.
The fastest tools don't help if people don't feel safe enough to use them. The prerequisites the AI mandate keeps skipping.
More from January
Workslop has a name. Readiness has a scorecard. Small AI has a case.
Get The Pepper Report
The short list of what we're writing and what we're reading.