3 ways AI agents will make your job unrecognizable in the next few years (www.zdnet.com)

Boomi CEO Steve Lucas predicts that AI agents will radically reshape the workplace within a few years: individual applications will be “consumed” by AI and persist only as back-end systems of record, while billions (even trillions) of lightweight agents will autonomously perform tasks via a natural-language “experience layer.” He illustrated this with a simple data-quality agent you can spin up in 30 seconds to clean addresses before they hit systems, and argued that routine actions (approving expense reports, checking calendars) will be handled proactively by agents so users may never log into traditional applications again. For the AI/ML community this accelerates two technical priorities. First, agent orchestration and agentic development platforms (Boomi, Bedrock, ServiceNow, etc.) must support secure, reliable connectors to legacy systems, workflow glue, observability and governance. Lucas calls that intermediary the AI “activation layer” — responsible for integration, policy enforcement, and ensuring ROI (MIT research shows 95% of enterprises struggle to get measurable results today). Second, research and engineering should focus less on single-model capabilities and more on multi-agent coordination, robust grounding to systems of record, data quality tooling, safety controls and monitoring to make these automated enterprises trustworthy and scalable.