Our Take
The honest assessment from operators at scale: current AI tools answer questions but rarely take action, and most enterprises lack the data infrastructure to change that.
Why it matters
HR leaders evaluating AI investments need realistic timelines for when tools will move beyond smart chatbots to actual workflow automation. Infrastructure constraints matter more than vendor promises.
Do this week
HR leaders: audit your system integrations and data governance before committing to AI agent pilots so you can avoid costly rollouts that can't deliver on automation promises.
Marriott says enterprise AI agents remain mostly chatbots
Kris Dunn, CHRO for Marriott's U.S. and Canada operations, told HR Tech Europe 2026 attendees that AI agents need to start taking action in 2027 because "it better be." Current AI tools largely function as smart chatbots that answer questions but rarely execute tasks, falling short of what most HR leaders were promised.
Francisca Martinez, Marriott's CHRO for Europe, Middle East and Africa, identified three barriers blocking meaningful AI agent deployment: poorly governed data connections, lack of real-time system integration, and organizational readiness gaps. At Marriott, despite having a global core HR system, dozens of standalone payroll and benefits platforms across EMEA cannot communicate with each other.
The executives, managing HR operations across $26 billion in revenue and 9,000 properties in 145 countries (company-reported), outlined a hybrid approach to HR technology that combines suite providers for compliance and scale with specialized tools for high-priority functions like talent acquisition.
Infrastructure beats vendor capabilities
Martinez warned that "the technology may arrive before the infrastructure does," highlighting a gap between AI vendor promises and enterprise reality. Even large organizations with substantial resources face data integration challenges that prevent AI tools from moving beyond question-answering to task execution.
The CHROs identified automated benefits changes, payroll questions, and absence management as prime automation targets, but cautioned against automating existing processes without first re-evaluating workflows that accumulated complexity over years.
Marriott's experience with adoption timelines provides concrete benchmarks: automated interview scheduling took 18 months to reach 80% adoption, while their manager scheduling tool runs at 70% adoption (company-reported).
Start small, fix data first
Martinez recommended AI-driven HR self-service as the fastest path to measurable ROI, delivering quick wins for both cost reduction and HR team reputation. High-volume transactions offer the clearest success metrics: faster employee responses and improved downstream data quality from associate-initiated transactions.
Dunn advised finding "one thing that's relatively contained" to prove technology value before expanding scope. The executives suggested treating employees as adoption tiers, providing selective access to advanced tools with clear expectations for knowledge transfer.
Marriott uses AI champions in India who receive broader tool access in exchange for educating colleagues and surfacing new use cases. This approach helps manage security concerns while building internal expertise before wider deployment.