Using AI vs Doing AI (Most Hospitality Players are Doing the Former)
There is a line Sol Rashidi drew at MURTEC 2026 that should be posted on the wall of every hospitality technology department in the industry. It is not a complicated line. But the gap between the two sides of it is where hundreds of millions of dollars in AI investment are quietly disappearing.
The line separates two fundamentally different organizational postures toward AI. Rashidi called them Using AI and Doing AI, and the distinction is not semantic. It is the difference between buying tools and building capability, and most hospitality organizations are firmly on the wrong side of it.
The Distinction That Changes Everything
Using AI means buying Copilot licenses, deploying chatbots on the booking engine, generating email summaries, running guest-facing virtual assistants. Plug-and-play. Minimal process change. Minimal ROI.
Doing AI is a fundamentally different undertaking. It is a strategic initiative requiring dedicated focus, sustained funding, and genuine business process redesign. It means breaking workflows into their component tasks and sub-tasks, assessing the risk profile of each, and making deliberate decisions about what can be delegated to AI and what must remain human-centered.
Most hospitality organizations are Using AI. They are buying tools. Vendors are selling licenses. And the productivity lifts promised in pitch decks are not materializing because the underlying business processes were never redesigned to capture them.
Rashidi was direct about why this happens. Doing AI is harder. It requires organizational commitment that Using AI does not. It requires executive sponsorship, cross-functional process ownership, change management investment, and the willingness to slow down before speeding up. For an industry already stretched thin on labor and operating margin, that kind of commitment is genuinely difficult to make.
The Sequence Problem
The most important principle Rashidi articulated in this section of her MURTEC keynote is one that the hospitality industry’s technology procurement culture consistently violates: true productivity gains require business process optimization first, then strategic AI integration.
In hospitality terms, this means fix the room assignment workflow, then automate it. Fix the labor scheduling logic, then let AI optimize it. Redesign the guest communication process, then deploy AI to execute it at scale. The sequence matters more than the technology.
What most operators are doing instead is layering AI on top of existing processes without interrogating whether those processes are worth automating. A labor scheduling process built around a head of department’s intuition and a spreadsheet does not become more effective when AI is applied to it. It becomes a faster version of the same flawed process, now with a technology budget attached.
Rashidi’s framing is precise: using AI on broken processes produces faster broken processes. The efficiency gain is real. The outcome improvement is not.
What Doing AI Actually Looks Like in Hospitality
The operational difference between Using AI and Doing AI becomes concrete when you apply it to specific hospitality functions.
A hotel group Using AI deploys an AI-generated response tool for guest communications. Staff receive suggested replies, approve them with minimal review, and response times drop. The metric looks good. The guest experience improvement is marginal because the underlying communication strategy was never examined.
A hotel group Doing AI maps every guest communication touchpoint, categorizes interactions by complexity and relationship sensitivity, identifies which categories can be fully automated, which require AI-assisted human response, and which require unassisted human judgment. It redesigns its staffing model accordingly, trains its teams on the new workflow, establishes quality review protocols, and then deploys AI into the process it has deliberately built to receive it. Response times drop and guest satisfaction scores move.
The technology deployed in both scenarios may be identical. The outcomes are not, because the organizational work that precedes the technology deployment is entirely different.
The Workforce Dimension
Rashidi connected the Using AI versus Doing AI distinction directly to workforce strategy, and the implications for hospitality are significant
Organizations that are Using AI are deploying tools on top of their existing workforce without meaningfully changing what their people do or how they do it. The productivity lift, where it exists at all, is marginal because the human work surrounding the AI tool was never redesigned.
Organizations that are Doing AI are asking a fundamentally different question: how do we exponentially amplify what we are doing with who we have, plus these capabilities, and potentially expand teams into new functions? That question produces different answers. It produces AI-augmented front desk teams who handle more complex guest needs. Along with that, it produces revenue management analysts who spend less time pulling data and more time building strategy. It also produces housekeeping supervisors with real-time AI-assisted dispatch who can manage larger floor plates without sacrificing quality.
The workforce that results from Doing AI is more capable, more engaged, and more valuable than the workforce that results from Using AI. It is also more expensive to build, which is precisely why so many organizations settle for Using AI and then wonder why the ROI never arrived.
The Vendor Accountability Gap
There is a structural problem in the hospitality AI vendor landscape that Rashidi’s framework exposes cleanly. Vendors are primarily incentivized to sell Using AI. Licenses, subscriptions, and implementation fees are attached to tool deployment, not to outcome achievement. The vendor’s revenue is not contingent on whether the operator redesigned their processes before deploying the tool.
This creates a market dynamic where the path of least resistance for both vendor and operator is Using AI. The vendor closes the deal. The operator reports an AI initiative to their board. The productivity gains do not materialize. Eighteen months later, the tool is quietly sunset and a new vendor begins the cycle again.
Breaking this cycle requires operators to demand more from the procurement process itself. The right questions before any AI deployment are not about the technology’s capabilities. They are about the operator’s process readiness. Has the workflow been mapped and optimized? Has the data been cleaned and governed? Have the humans in the workflow been repositioned as editors and validators rather than passive recipients of AI output? If the answers are no, the deployment is Using AI regardless of how sophisticated the technology.
The Bottom Line
The hospitality industry does not lack access to AI tools. It lacks the organizational discipline to deploy them in a way that produces durable results.
Rashidi’s distinction between Using AI and Doing AI is not a criticism of the tools available. It is a diagnosis of the organizational posture required to extract value from them. The tools are largely sufficient. The processes surrounding them are largely not.
Every hospitality operator currently evaluating an AI investment should ask one question before signing: are we prepared to redesign the process this tool will support, or are we buying a faster version of what we already have? The honest answer to that question determines whether the investment will appear in the next board presentation as a success story or as a line item in the POC graveyard.
Up Next in the Series:
This was Post 3. Post 4 examines the hospital case study Sol Rashidi shared at MURTEC 2026 and what its lessons about workflow design mean for every hospitality operator deploying AI-generated guest communications, revenue recommendations, or demand forecasts.
IHL Group covers retail and hospitality technology markets globally. For more information on our research, visit https://www.ihlservices.com. Sol Rashidi keynoted MURTEC 2026 in Las Vegas. All data and frameworks cited in this post are attributed directly to her presentation.