The Meeting With No Memory

An educational briefing on applying generative AI to the hotel's daily operations meeting — cross-referencing “The 7% Problem” and HBR Guide to Generative AI for Teams (Farri & Rosani, Capgemini Invent, Harvard Business Review Press, 2026)

Walk into almost any hotel on earth at the same hour every morning, and you will find the same meeting. The GM chairs it. The operational heads — front office, housekeeping, F&B, engineering, security, sales, sometimes revenue — sit around the same table or the same Teams grid. The agenda never changes: what happened last night (the post-mortem — incidents, complaints, no-shows, near misses, a VIP who wasn't happy), what's happening today (arrivals, departures, groups, events, staffing, weather), and a glance ahead at the next few days. Decisions and actions then cascade down through each department head to the shift briefings and the line staff who actually deliver the day.

It happens 365 times a year, in every property, forever — and per the Capgemini Research Institute survey behind Farri and Rosani's HBR Guide to Generative AI for Teams, it sits almost entirely inside the 93% of team meetings that still run without AI in the room (see “The 7% Problem”). This is the piece on what changing that would actually look like, chapter and page references included.

TL;DR

• The morning briefing is a daily version of what the HBR Guide calls a Business Review — and it inherits that activity's exact failure mode: “a person from each function reads out numbers already visible on dashboards, and the meeting becomes an information broadcast” (ch.26, p.271).

• Medicine's own version of this meeting — the Morbidity and Mortality (M&M) conference — was built specifically to fix that: blame-free, systemic, focused on why something happened, not just what. A century of practice shows it drifts back into recitation whenever nobody designs against that pull.

• AI's job in a hotel morning briefing is not to replace the GM's judgment. It's to do the unpaid labor that currently eats the first ten minutes (assembling the numbers) and the unpaid labor that currently never happens at all (spotting the pattern across five departments' worth of yesterday).

Bottom line: Build it to synthesize before, challenge during, and cascade after — or it's just a status report with better formatting.

A borrowed discipline: what M&M gets right, and how briefings drift

The Morbidity and Mortality conference has been part of medicine for over a century — tracing back to Ernest Codman's push to track surgical outcomes in the early 1900s, formalized after the 1910 Flexner Report and the founding of the American College of Surgeons in 1912. The idea was, and still is, simple: gather the team regularly, review what went wrong, and fix the system — not the individual. Modern guidance describes the standard well: discussions should be blame-free, multidisciplinary, and focused on education and systemic improvement rather than assigning fault.

Here's the uncomfortable part, and the reason this analogy is useful rather than flattering: M&M conferences fail constantly, in a very specific way. Research reviewing hospital practice found that in one study, 72% of cases discussed involved neither real morbidity nor mortality — padding, not learning — and that across four major hospitals, “most of the allotted time was spent on case presentation and guest speaker commentary, with very little audience participation or discussion of error.” Even the meeting built to prevent recitation has to keep fighting the pull toward recitation.

That is precisely the pull the HBR Guide describes in the hotel morning briefing's nearest formal cousin, the Business Review — typically monthly at business-line level, sometimes quarterly across divisions (ch.26, p.271). Left undesigned, “each function reports in isolation, and nobody connects the dots between sales, operations, R&D, and HR” (p. 271) — replace those four functions with front office, housekeeping, F&B, and engineering, and you have described this morning's briefing at more hotels than not. A hotel morning briefing run daily inherits that failure mode at 365x the frequency of a monthly business review, which is exactly why it's worth fixing properly rather than living with it.

Before: give the room a memory

The first ten minutes of most morning briefings are spent assembling the numbers — someone reading the overnight log, someone else finding last night's incident report, a third person checking today's arrivals in the PMS. None of that is judgment. All of it is exactly the kind of synthesis the HBR Guide assigns to the “Before” phase of a review meeting: “Gen AI doesn't just analyze KPIs or flag trends — it can help teams make sense of data. This reduces information overload and ensures that attention is focused on the meaning behind the data, not merely the numbers themselves” (ch.26, p.272).

Applied to a hotel, this means one digest waiting on the table before the GM sits down — built from the PMS (arrivals, departures, VIPs, groups, no-shows), the overnight incident and engineering logs, guest complaint tickets, security log, occupancy and rate position, and tomorrow's weather and local events. Not a report to read aloud. A set of questions to argue about. Adapted from the book's own “focus discussion questions” and “anticipate discussion dynamics” prompts (ch.26, pp.274–275), a hotel version might ask the model to:

• Summarize last night's incident and complaint log, and flag anything that touches more than one department — that's the pattern a single department head can't see from their own log alone.

• Given today's arrivals, departures, groups, and events, name the three staffing or service risks most likely to produce a guest-facing failure today, and which department owns each one.

• Compare this week's recurring issues against last week's. What did we say we'd fix that's still showing up?

None of this is exotic. It's the same “synthesize performance, focus discussion questions, anticipate discussion dynamics” structure the book sets out for a monthly Business Review (ch.26, pp.272–275), run every single morning instead, because that's the actual cadence of a hotel.

During: make it a real M&M, not a recitation

This is where discipline matters most, because both source traditions warn about the exact same failure. The HBR Guide's Root Cause Analysis chapter names it directly: teams under pressure “confuse symptoms with causes, stopping at surface-level explanations like ‘operator error’ or ‘software bug’ without asking why those occurred,” and “in cultures where people fear blame, they may withhold information, keeping the true causes hidden” (ch.23, p.242). That is the hotel version of the M&M literature's own warning: that discussions must stay “blame-free” if they're going to surface systemic issues rather than just naming a scapegoat.

A well-designed AI layer earns its place here by doing the thing a tired department head at 8am often won't: asking the second question. Three separate air-conditioning complaints from three different guests on the same floor isn't three isolated grumbles — it's an engineering pattern, and it should be named as one before the meeting moves on, not discovered separately by three department heads over the following week. This is also exactly where the three conditions from “The 7% Problem” apply without adjustment: intentionality (someone has to decide this meeting will run this way, deliberately, not default to how it's always been run), craft (a digest that only reads out numbers has automated the recitation, not fixed it), and collective ownership (the moment department heads start nodding along to what the AI summary said instead of arguing about it, the meeting has quietly stopped being an M&M and gone back to being a broadcast).

If your team is just watching the screen, that's not a good sign — AI is taking the lead, not the team.

from “The 7% Problem,” on the same warning sign applied here

After: closing the loop down the chain

The morning briefing's output has to survive three or four more relays before it reaches the housekeeping attendant or the front desk agent actually on shift — GM to department head, department head to supervisor, supervisor to line staff — and every relay is a chance for it to be diluted, garbled, or forgotten by lunchtime. The HBR Guide's fix for this, again written for a monthly review but suited even better to a daily one, is to stop sending one recap to everyone: “Instead of sending the same meeting recap to all participants, gen AI helps produce tailored summaries for each person or function based on their role, contributions, and responsibilities” (ch.26, p.279).

For a hotel, that means the housekeeping supervisor's version of this morning's briefing looks nothing like the F&B captain's version — each gets the two or three lines that are actually theirs to act on, in the order that matters to their shift, not a transcript of a meeting they weren't in. The other half of “after” is less glamorous and more important: checking whether yesterday's action items actually happened. A morning briefing that raises the same unresolved issue for the fourth day running, without anyone noticing it's the fourth day, has failed at the one thing M&M conferences exist to do — “modify behavior and judgment based on previous experiences, and prevent repetition.”

THE PERTLINK VIEW

None of this replaces the GM's judgment about which guest issue actually matters, or the department head's read on whether a pattern is a fluke or a trend — that's the {HXO} Human Experience Orchestrator's job, not the model's. What AI changes is what the room is arguing about by the time the coffee's poured: not whose numbers are missing, but what the numbers actually mean.

This is also, not incidentally, exactly the kind of AI use case that should come first, well ahead of the flashier agentic pilots competing for the same budget: cheap, high-frequency, low-drama, and running every single day of the year whether or not anyone's watching. Get the Token Cost Per Guest math right on 365 short daily briefings before spending it on the initiative that only runs once.

The intelligence may be artificial. But the experience is human.

Made with the help of various AI tools – but always with a HITL

Sources

• Farri, E. and Rosani, G., HBR Guide to Generative AI for Teams (Boston: Harvard Business Review Press, 2026) — Chapter 23, pp.241–242; Chapter 26, pp.271–281.

• “The 7% Problem” — Pertlink Viewpoint, on Farri and Rosani's HBR Press Live launch webinar and the Twin Benefit / Three Conditions framework for team AI adoption.

• Pierluissi, E. et al., cross-sectional review of morbidity and mortality conferences at four major hospitals, cited in “Transforming the Morbidity and Mortality Conference into an Instrument for Systemwide Improvement,” AHRQ Advances in Patient Safety, Vol. 2.

• “Morbidity and Mortality Conferences: A Mini Review and Illustrated Application in Veterinary Medicine,” PMC5845710; “Morbidity and mortality conference,” Wikipedia; Clinical Excellence Commission (NSW), “Morbidity and mortality meetings.”