This podcast episode explores the current state and future of AI agents in the workplace. Jonas Dietzun, CEO of Beam AI, discusses how self-learning AI agents are being deployed in HR and finance, often with 95%+ accuracy, and explains the shift from simple agents to complex 'digital twins' of entire companies. He offers a cautiously optimistic view of the future, arguing that AI will change jobs rather than eliminate them entirely, while also introducing the concept of AI systems simulating business decisions before they are made.
Summarized by Podsumo
AI agents can now achieve over 95% accuracy on complex business tasks, especially in finance and HR, where they are used for invoice reconciliation, sales order processing, and screening resumes.
Jonas introduces 'digital twins' of entire companies as the next evolution beyond AI agents, where AI learns to simulate the entire organization and make autonomous decisions.
Despite fears of job losses, Jonas argues that AI often increases productivity, leading to *more* work being generated, not less, as people want more of what the AI can produce.
The biggest challenge is not the technology itself but the organizational 'implicit knowledge' that AI cannot access, a problem that is being solved by observing inputs and outputs.
The most exciting development is the move towards 'fully autonomous companies' where AI orchestrates entire business operations, although this is still a complex and evolving challenge.
"If I now look at the history, 2025 was about experimentation, and now 2026 we really see agents being put into production and live, and then having the benefit."
"I see the next step as building complete systems in which AI agents operate. We are building a digital twin of the company."
"I know nobody in AI who works less because of it. If AI were really helping to cut jobs, I would see people working less. But they don't."