How to read the evidence
Guest-memory decisions involve several different questions. Does personal service matter? Will staff record useful knowledge? Does it reach another colleague? Does acting on it improve the guest’s experience? Does that experience eventually lead to another booking? One source rarely answers them all.
This selected research register connects published findings with AFG’s operating proposition. It is an original interpretation, not an exhaustive review of every hospitality article. Independent sources establish the wider context. AFG’s own records describe its implementation. An operator trial is needed to assess results in a new setting.
1. Guest data, governance and the human contribution
Industry report · PwC and NYU · 2023
The report Hotel industry digital transformation: The current state of play draws on industry interviews and focus groups. It links guest knowledge with personal service while emphasising controls, integration and the people needed to make technology useful.
AFG’s interpretation: a guest-memory proposition should explain how staff supply useful knowledge and how the organisation governs its use. Collecting more data alone does not complete that task.
Scope: this is industry research, not a test or endorsement of GuestMemoryOS. Read the original report, particularly pages 3 and 18.
2. Service matters to the return decision
Hospitality research · Gallup · 29 July 2014
Gallup’s hospitality analysis associates stronger engagement with higher spending and lower price sensitivity. It identifies responsive employees, reliable service and problem resolution as relevant to repeat-booking decisions, with differences between generations.
AFG’s interpretation: test whether retained service context helps employees respond and follow through. A record that never affects service cannot be assumed to create loyalty.
Scope: these historical findings do not provide a current revenue forecast or a percentage uplift attributable to guest-memory software. Read Gallup’s original analysis.
3. The guest’s remembered experience and future intentions
Peer-reviewed research · Jong Hyeong Kim · 2018
Kim’s Journal of Travel Research study examines relationships between memorable tourism experiences, destination image, satisfaction, revisit intention and word of mouth. Its reported model connects memorable experiences with future behavioural intentions.
AFG’s interpretation: distinguish the organisation’s stored memory from the experience the guest remembers. Software stores context; people use it to deliver hospitality.
Scope: the university-hosted abstract and bibliographic record were reviewed. The research concerns tourism experiences and intentions; it is not a GuestMemoryOS trial or a measurement of completed repeat bookings. Read the abstract and publication details.
4. AI requires evaluation and risk management
Voluntary framework · NIST · AI RMF 1.0, 2023
NIST’s AI Risk Management Framework supports organisations in managing AI-related risks. It provides a general foundation for assessing how AI is used, rather than declaring a particular hospitality implementation safe or accurate.
AFG’s interpretation: retain the distinction between a source observation, a proposed interpretation and a prediction. Assess the actual workflow and its human decisions in the intended setting.
Scope: citing the framework is not a claim of NIST certification or an independent audit of GuestMemoryOS. Read NIST’s framework overview.
5. What AFG’s live record adds
The AFG operating-evidence page records a live residence deployment at Mandarin Beach Villa. It separates the activity window of 14 July–21 September 2026 from the property’s review record. The published activity snapshot includes 5,375 staff notes and 114 recorded actions; those are different event types, not interchangeable measures of guest benefit.
The staff adoption account reports three days of training and continued contributions after the trainer’s departure. AFG also reports an average QuickNote entry time of six seconds. That measures note entry, not total observation time, management review or daily workload.
The 40 five-star Airbnb reviews are a property-level outcome checked on 21 September 2026. They do not isolate the software’s effect. Internal cruise and hotel simulations are a separate evidence category from live use. No new independent audit or causal performance study is represented by this register.
One family booking.
Every guest remembered individually.
Explore AFG’s reported 40-family trial setting, Mandarin Beach Villa’s public reviews and what person-specific memory could mean for five cruise operating models.
Company-reported family cohort · Public review context · Proposed cruise evaluations
Read the multi-generational guest identity paper →Further sources for the reference series
The new series adds primary guidance on provenance, accuracy and evaluation. These sources inform AFG’s proposed methods; none is a GuestMemoryOS product endorsement.
- W3C (2013). PROV-Overview: An Overview of the PROV Family of Documents.
- Working Group Note, 30 April 2013; abstract and introduction. General provenance concepts. Reviewed 25 September 2026.
- Information Commissioner’s Office. Principle (d): Accuracy.
- Official UK guidance; sections on opinions, sources and keeping information current. Jurisdiction-specific guidance. Reviewed 25 September 2026.
- HM Treasury. Guidance on the Impact Evaluation of AI Interventions.
- Official evaluation guidance; choosing an evaluation approach. Applied here to hospitality by AFG. Reviewed 25 September 2026.
- HM Treasury. Magenta Book Annex A: Analytical methods for use within an evaluation.
- Official guidance; section A2.7, difference-in-difference. Methodological reference. Reviewed 25 September 2026.
See the reference series for the claim-level citations and hospitality applications.
Airline primary sources and new evidence sheets.
Original analysis of documented airline service models, connected to AFG’s guest-memory evidence and operating methods.
Primary sources for the aviation collection
The aviation collection adds operator documentation from Emirates, Qatar Airways and Singapore Airlines, together with IATA ONE Order and Travelport service-request guidance. The articles distinguish existing capabilities from AFG’s proposed incremental contribution. Airline sources support operating-model facts, not GuestMemoryOS outcomes.
The claim-to-evidence map connects these applications to the dated AFG record; the record specification applies general provenance and evaluation principles to service.
AI needs a method.
Guest memory needs evidence.
Read how AFG’s reported experience of repeated failures during live development shaped the GuestMemoryOS approach—and why identity, context, corrections and service follow-through matter.
Why AI alone is not guest memory →AFG live-development account: AI and methodology
The live-development methodology paper records AFG’s 25 September 2026 account of repeated incorrect information during AI-led attempts and the development of its operating method through live trials. This is a qualitative company source. Its NIST, LongMemEval and W3C references support general concepts; they do not validate AFG’s results or establish exclusive effectiveness.
From a cited source to a testable claim
A useful claim has a short chain: source → what it establishes → operating implication → trial measure. For example, service responsiveness is relevant to repeat visits; AFG proposes passing useful context to the next colleague; the trial measures whether that colleague receives and uses it. Actual return bookings require later observation.
Internal links in this library explain related concepts; they are not extra independent corroboration. Research findings, company reports and illustrative scenarios retain their own labels throughout.
Apply the trial scorecard · Read the GuestMemoryOS selection case