The perennial Superbowl of Hotel Tech was a whirlwind this past June in San Antonio. Going in, we all wanted AI hard and fast, with the shift happening clear as day that the tools on display on the expo floor were more than mere chatbots like past outings.
The real story, though, is agent management or dare we use a shnoozefest word: governance. As hotels begin deploying dozens of AI-powered workflows across revenue management, marketing, reservations, maintenance, finance and guest communications, the challenge is taking stock and reining in all these agents. In this sense, HITEC 2026 felt like the moment the industry began realizing that AI agents are becoming operational infrastructure, which screams for more transparent oversight.
The Rise of the Agent Management Platform
The most important emerging software category I observed at the show is what I would call the Agent Management Platform (AMP). Think of it as the governance layer for an organization’s AI agent ecosystem.
Today, a hotel may have one workflow using a frontier model, another powered by a small language model (SLM) and a third embedded within a vendor platform. Each agent accesses different data, operates under different permissions and incurs different costs. That complexity compounds rapidly.
Which models are being used? What guest data can they access? What actions are they authorized to take? What happens if an agent fails? Is there an audit trail? Can personally identifiable information be scrubbed before processing?
These questions are important today. They become mission-critical once hundreds of agents are operating simultaneously across a hotel portfolio. The AMP becomes the control center that governs permissions, monitoring, security, performance and cost.
Token Creep Is Coming
Token costs are about to become a very real part of the tech stack opex budget. At present, AI remains relatively inexpensive compared to labor, distribution costs or technology licensing. Many organizations treat token consumption as an afterthought. These things have a tendency to creep upward once embedded.
As AI adoption accelerates, hotels will discover that thousands of automated decisions, guest interactions, forecasts, reports and content generation tasks create a new operating expense category. The industry will eventually scrutinize token consumption the same way it analyzes labor costs, OTA commissions or payment processing fees. Expect that ‘token spend optimization’ will become a standard management discipline within a few years.
Not every task requires the most expensive model available. A simple guest inquiry may only need a lightweight, low-cost model. A strategic pricing recommendation or investment analysis may justify a more sophisticated system.
The ability to route tasks intelligently based on performance requirements and cost considerations will become a competitive advantage. This is where the AMP becomes the hub. Beyond governance, it becomes a budgeting tool that continuously arbitrages AI workloads amongst different models to maximize return on every token consumed.
Managers Must Learn Loop Engineering
Loop engineering is the design of self-improving operational cycles where AI observes, recommends, acts and learns from outcomes. Similarly, harness engineering is the discipline of constraining and governing those systems so they remain aligned with business objectives.
For hotel leaders, this represents a profound shift in how work gets done. Rather than personally managing every process, managers increasingly become architects of decision systems, defining objectives, establishing guardrails and determining where human judgment creates the most value.
The difficult part is not the technology, though. It is having the confidence to trust well-designed loops while maintaining accountability for the outcome.
And that’s the real lesson from HITEC 2026. To realize the digital transformation oozing from this tradeshow, a hotel organization needs to take bold steps to rethink their business processes in order to transition from the Internet Age to the Age of AI.
The winners will not necessarily be the organizations deploying the most agents. They will be the ones that know exactly which agents are operating, what they are allowed to do, how much they cost and how effectively they contribute to business performance.
