Misryoum looks at how AI agents are being used for real work, and what that means for jobs and productivity.
An “AI taking jobs” narrative is spreading fast, but one practical experiment suggests the story is more complicated.
In this case. Misryoum reports that BI’s Amanda Hoover built an AI bot trained on her own body of work to see whether it could actually perform her day-to-day tasks.. The test went beyond generic automation: the bot was used to handle activities ranging from conducting interviews using a voice agent to drafting and coordinating work with an editor.. For Hoover, the takeaway was not a simple alarm bell.. Instead. the exercise functioned like a reality check on what current systems can do well. and where they still fall short.
Misryoum insight: When AI is trained on a real workflow, the question shifts from “Can it write?” to “Can it execute responsibly under deadlines and approval processes?”
Meanwhile, Misryoum also highlights a different style of deployment inside companies: orchestration rather than one-off tasks.. Andy Cabasso. who manages dozens of AI agents for a productivity platform. describes using multiple agents to cover functions such as pulling analytics and scheduling follow-up meetings.. In his setup. agents are even given distinct fictional personas and tailored negotiation behavior for specific use cases. reflecting a growing trend toward treating AI as an operational workforce.
This approach signals a shift in how organizations think about AI.. Instead of asking tools to perform isolated tasks. teams are increasingly setting goals and letting systems reason through steps to reach outcomes.. That change matters because it can compress time spent on coordination. but it also raises the stakes of how well the system understands context and how consistently it follows internal standards.
Misryoum insight: Letting an AI system “decide the steps” can boost speed, but it also makes quality control and oversight a core part of the job rather than an afterthought.
Misryoum notes that how far companies push these tools may depend on the role and seniority of the human involved.. Executives with large, relentless schedules may be more willing to delegate work to AI systems.. Less experienced employees, by contrast, often still need time, feedback, and learning loops.. That difference creates a practical tension: will leaders wait for juniors to build competence. or will they replace waiting time with bot-driven output?
Ultimately, Misryoum frames the broader question in workplace terms.. The future likely won’t be a straight line from AI to job loss.. It may look more like job redesign. where humans manage objectives. review results. and handle the edge cases that automation struggles with.. The real test. in other words. is not whether AI can do work. but whether organizations can integrate it without hollowing out the skills and judgment that keep work reliable.
Misryoum insight: The most durable advantage may shift toward workers who can supervise AI effectively, translate goals into clear instructions, and recognize when a system is likely to be wrong.