AI Agents & the Future of Work | Forkast

Agents at Work

AI Is Devaluing Specific Tasks Within Jobs. Payroll Data Now Proves It.

A landmark study from ADP Research and Stanford maps which IT tasks employers are discounting as AI spreads — and it explains why enterprise execution failures keep climbing.

◆ Dana EllisonJul 31

Agents at Work

Enterprise AI Failure Modes Have Shifted. Hallucinations Are No Longer the Problem.

New research from ChatSee.ai analyzing 10,000+ failure events shows hallucinations now account for less than 10% of enterprise AI failures. Execution and escalation breakdowns dominate — and budget allocation hasn't caught up.

◆ Dana EllisonJul 29

Agents at Work

MCP Locked In Its Architecture Today. Now the Hard Part: Enterprise Adoption at Scale

41% of software orgs say they have MCP in production. The data behind that number tells a different story.

◆ Dana EllisonJul 28

Agents at Work

The AI Agent Deployment Gap: 80% Embed, 31% Ship — and Workers Pay the Price

The announcement curve is vertical. The deployment curve is a slow climb. The binding constraints are organizational, not technical — and the cost falls on workers waiting for tools that never arrive.

◆ Dana EllisonJul 28

Agents at Work

Cognition’s Devin Writes 89% of Its Own Company’s Code. The Real Story Is What’s Left for Humans.

The autonomous software engineer has evolved into a $26B full-lifecycle platform – writing code, finding vulnerabilities, running production systems. But the babysitting tax of managing AI that confidently does the wrong thing is the hidden cost nobody's counting.

◆ Dana EllisonJul 22

Agents at Work

The EU AI Act Employer Deadline You Were Counting Down To Just Split in Two

High-risk employment AI rules got pushed 16 months. Transparency obligations didn't. Seventy-eight percent of organizations haven't started preparing. The compliance timeline is now the competitive variable.

◆ Dana EllisonJul 21

Agents at Work

Meta’s AI Tracked Their Keystrokes. Then It Put Them on the Layoff List.

Twenty-six employees say Meta's productivity-monitoring AI penalized them for taking legally protected leave. Separations proceed tomorrow. This is the first major test of whether that's discrimination.

◆ Dana EllisonJul 21

Agents at Work

The Attribution Gap: Why Employers and Workers Can’t Agree on What Counts as an ‘AI Layoff’

Employers cite AI when it justifies layoffs and disclaim it when it invites scrutiny. Workers carry the cost in the gap between those two accounts — and nobody is tracking the difference.

◆ Dana EllisonJul 20

Agents at Work

Enterprises Are Running AI Agents in Production. Most Can’t Control Them.

New data shows 65% of organizations have experienced an agent-caused security incident, 60% can't terminate a misbehaving agent, and 63% can't enforce purpose limits. The gap isn't deployment — it's governance.

◆ Dana EllisonJul 18

Agents at Work

85% of Enterprises Are Testing AI Agents. Only 5% Are Running Them. The Problem Isn’t Capability.

Cisco data shows a massive gap between agent experimentation and production deployment — 85% of enterprises piloting AI agents, only 5% in production. Amazon's AGI director and Princeton researchers say reliability, not intelligence, is the binding constraint.

◆ Dana EllisonJul 17