Preserving the Human: How The Inn at Little Washington Is Exploring AI as a New Medium for Hospitality, Culture, and Legacy

A Forty-Year Hospitality Institution Experiments With Its Next Chapter

At The Inn at Little Washington, the biggest question surrounding artificial intelligence isn't around what it can automate, but what might it help preserve.

Since opening in 1978, The Inn at Little Washington has built an institution around an unusually expansive idea of hospitality. Food is essential, but the experience extends well beyond the plate. Language, anticipation, memory, theatricality, place, and a near-obsessive attention to the individual guest are woven into something deliberately difficult to standardize. At the center of it is Chef Patrick O'Connell, and the philosophy, vocabulary, stories, and creative instincts he has developed over decades.

For Flatter, Inc., the relationship with the Inn started with a personal connection rather than a sales opportunity. CEO Brittany Chiang describes founder Dr. J.R. Flatter as a longtime patron, fascinated for years by the Inn's business model and approach to hospitality. So when the Flatter team began developing PersonaOPPs, the company's conversational AI platform, Chef Patrick O'Connell was a natural subject for an early experiment in what a digital persona could become.

In 2025, around O’Connell’s 80th birthday, members of the Flatter team brought an early digital-twin persona to Washington, Virginia. The experiment became affectionately known as “Little Patrick.” According to Chiang, the demonstration prompted a broader discussion with the Inn’s leadership about what artificial intelligence could mean for the business—not simply as another technology system, but as a new medium through which the Inn might explore its history, culture, and guest experience.

That discussion evolved into a formal technology and program-management engagement. The resulting scope organized Flatter's continued work around three connected workstreams: vision and experience design, IT architecture and security, along with archiving and digitization. These workstreams are supported through project planning, governance, stakeholder coordination, reporting, and oversight.

The program ambition, therefore, is considerably larger than building an AI avatar. Rather, it is an enterprise experiment regarding how technology can help a world-class institution preserve not only what is distinctive about itself, but also create new ways for patrons from across the globe experience the uniqueness as well.

The Challenge: How Do You Scale Something Intentionally Human?

Organizations are generally good at documenting explicit knowledge. Procedures become manuals. Policies become repositories. Training becomes courses. Historical materials become archives.

The harder knowledge to preserve is often the knowledge that matters most.

How does an exceptional maître d’ recognize that a guest wants attention without asking for it? How does a leader communicate a standard without reciting a policy? How does a founder tell a story in a way that teaches employees what the organization values? How does someone learn not merely what an institution does, but how that institution thinks?

In hospitality, these questions become especially consequential because the product is partly human interaction itself.

The Inn’s challenge is therefore an interesting one: its culture has been cultivated precisely because hospitality cannot be reduced to a transactional script. The proposal underlying the Flatter engagement recognizes this tension. It describes the Inn as combining “timeless craftsmanship and forward-looking vision” and argues that its digital interactions should ultimately aspire to the same level of intentionality as its physical hospitality.

This reframes organizational culture as a form of intellectual capital.

Patrick O’Connell’s knowledge exists partly in recipes, records and documented history. But it also exists in language, anecdotes, judgments, turns of phrase, standards, humor and the philosophy through which he interprets hospitality. Chiang describes Little Patrick as incorporating O’Connell’s actual voice and distinctive lexicon, including the whimsical vocabulary associated with his communication style. The intention is not to substitute an artificial persona for the person. It is to explore another way of making elements of that knowledge accessible.

For employees, that could eventually mean encountering the philosophy behind the standard rather than simply receiving the standard.

For an enterprise, that distinction is profound.

The Experiment: From Patrick O’Connell to “Little Patrick”

PersonaOPPs is designed around a simple construct: an OPP combines a Persona, a defined Scenario or Case Study, and an Analysis or Output.

That structure matters. PersonaOPPs is not intended to be an unrestricted chatbot that happens to sound like a particular person. The underlying business concept is a bounded conversational experience designed for a particular purpose. A persona provides character and context; a scenario creates the reason for the interaction; and the resulting transcript or structured analysis turns the conversation into something that can be reviewed, learned from, or incorporated into a broader process.

The platform combines Google Gemini for conversational intelligence, ElevenLabs voice capabilities, and an AWS serverless cloud architecture, with configurable personas, source knowledge, scenarios, transcripts and structured outputs. Human design and governance surround the technology: what the persona should know, how it should behave, what source materials are appropriate, what an interaction is intended to accomplish, and where humans remain responsible for interpretation and decisions.

Little Patrick takes this model into different territory.

The Inn’s formal vision calls for exploring the persona across three distinct audiences: internal employees, guests, and potentially Smithsonian visitors. The work includes an employee journey for training and help-desk access, a guest journey extending from pre-arrival through post-visit, potential educational and thought-leadership experiences, and a Smithsonian-oriented concept blending heritage and technology. The stated goal is unusually revealing: to tell the Inn’s story through “experience, education, and preservation” in Patrick O’Connell’s voice.

That is fundamentally a knowledge-management proposition.

Instead of asking someone to search an archive for Patrick O’Connell, the longer-term possibility is to allow someone to converse with carefully curated elements of the archive through a purpose-built persona.

That distinction also makes the Inn’s archival initiative strategically important.

The engagement calls for developing a searchable digital archive of historical materials including menus, letters, guest lists and photographs, along with an intake-to-digitization workflow, common metadata language, rights and permissions management, and an initial 100-artifact demonstration archive. The objective is explicitly to create a “living digital archive” capable of supporting future exhibits, media and training.

Together, the archive and conversational layer suggest an intriguing evolution in institutional memory:

First preserve the knowledge. Then organize it. Then consider how people might interact with it.

A Memoir, a Digital Persona, and Two Different Forms of Legacy

The timing adds another dimension.

In Chiang’s account of the project, Patrick O’Connell’s 2026 memoir, Maine, Middle and Gay, is important because it represents another effort to document a life, perspective and creative identity. She sees the richness of that source material as potentially contributing to Little Patrick, subject to the Inn’s decisions about appropriate content and use.

A memoir and an AI persona are radically different forms.

A memoir fixes stories into a deliberate narrative. A conversational persona raises the possibility that selected elements of an individual’s documented knowledge could instead be encountered interactively.

That does not make a digital persona a replacement for its subject. Nor should it.

The more compelling idea is that organizations may eventually have choices about how institutional memory is experienced. Some knowledge belongs in books. Some belongs in archives. Some is best transmitted through human mentorship. Some may be appropriate for structured conversational systems.

Little Patrick sits at the intersection of those possibilities.

Hospitality as an Innovation Laboratory

The Inn provides an unusually demanding environment in which to test the proposition because poorly applied technology could easily undermine the very experience it is intended to support.

The standard cannot simply be efficiency.

It must be hospitality.

Chiang uses the Inn’s concept of “dream weaving” to describe the larger opportunity: understanding guests well enough to create moments that feel unusually personal, imaginative and considered. In her account, ideas under exploration range from staff development and archival interaction to research involving food and allergies, virtual exploration of the grounds, and knowledge associated with the Inn’s herb garden. These are possibilities being investigated, not claims of deployed functionality.

The formal technology roadmap similarly contemplates potential guest-facing experiences while recognizing the infrastructure required to support them responsibly. It includes privacy and VIP protection, identity and network security, AI content governance, and a target-state architecture.

Other concepts are surprisingly tangible. The proposal explores outdoor points of interest paired with discreet digital interactions that could provide curated historical context, as well as multilingual support and potential physical endpoints for guest interaction with Little Patrick.

Another concept is a unified guest identity layer capable of connecting dining interactions, lodging and spa history, preferences, celebrations, and communications. Importantly, the proposal describes such a system as being used for “identity, memory, and messaging” rather than inventory control.

That phrase captures much of the opportunity.

Hospitality organizations have always depended on memory. The technological question is whether systems can help employees remember more—while preserving the discretion, judgment and humanity required to know what to do with that information.

Beyond Little Patrick: Building the Conditions for Responsible Experimentation

Flatter’s role consequently extends beyond PersonaOPPs.

The engagement establishes Flatter as a PMO and specialized information-systems advisory arm for projects outside the Inn’s day-to-day luxury-hospitality operations. Its remit includes digital information projects, archival work, potential AI integration into business processes and customer touchpoints, technology strategy and execution.

The work described by Chiang has included custom development, consulting, surveys, interviews, education and a cybersecurity maturity review. The formal scope adds governance, milestone management, architecture and security planning, archive design, stakeholder coordination and content governance.

This reflects a broader lesson about enterprise AI adoption.

Access to a sophisticated model is increasingly easy. Turning that capability into something an institution should actually use is considerably harder.

It requires discovering worthwhile problems before building solutions. It requires understanding data and institutional knowledge. It requires architecture and cybersecurity. It requires determining who should interact with a system and why. It requires workforce education and governance. And it requires measurement.

The Inn and Flatter have already identified success metrics as part of the intended roadmap, but the supplied materials do not yet establish quantitative ROI or performance outcomes for Little Patrick. Future evaluation should therefore examine measures such as employee adoption, frequency of voluntary practice, knowledge retention, training-time savings, employee confidence, guest engagement, usefulness of archival interactions, response quality and—perhaps most importantly for this institution—whether users believe the experience authentically reflects the Inn’s standards and culture.

What a Global Enterprise Can Learn From a Small Inn in Virginia

The Inn at Little Washington may seem an unlikely case study for a large federal enterprise.

That is precisely what makes it useful.

The Department of the Air Force, intelligence organizations and other large institutions possess a version of the same problem at radically different scale. Critical knowledge is concentrated in senior leaders, experienced operators, instructors, maintainers, acquisition professionals, technical experts and people approaching retirement. Some of that knowledge is documented. Much of it is not.

And the scarce resource is often not information.

It is access to experienced people, repeatedly, at the moment someone needs to learn.

The traditional answer has been to bring learners to experts: classrooms, mentoring, role-play, exercises and apprenticeship. Those approaches remain valuable. But they are inherently constrained by human availability.

Conversational AI introduces another possibility. Once appropriate knowledge is captured and governed, purpose-built personas could potentially allow people to encounter elements of expert knowledge asynchronously. Scenario-based systems could allow learners to rehearse difficult conversations, interviews, leadership situations or other consequential interactions repeatedly without requiring a second human being for every repetition.

The Inn offers an important warning alongside that opportunity: do not begin with the technology. Begin with what is precious.

What knowledge would hurt the organization if it disappeared? What judgment takes years to acquire? What behaviors distinguish merely competent performance from excellent performance? Which parts can be documented? Which must remain human? And which could be practiced through simulation?

Flatter has already applied PersonaOPPs in federal learning environments; its proposal cites use of the platform in Space Force training and Department of the Air Force senior-leader education. The Inn extends the underlying thesis into a very different domain and, in doing so, makes the idea easier to see.

A chef's philosophy of hospitality and a senior military leader's accumulated judgment are not equivalent knowledge.

But both demonstrate the extraordinary amount of organizational value that can reside inside people.

Looking Forward: Using Technology to Preserve the Human

The most compelling future for Little Patrick may ultimately have less to do with creating a convincing digital Patrick O’Connell than with what the experiment teaches the Inn about itself.

To build such a persona thoughtfully, an organization first has to ask what makes Patrick's perspective distinctive. It has to examine his language, stories, philosophy and standards. To create an archive, it must decide which artifacts matter and how they relate. To create scenarios, it must articulate what excellent hospitality looks like in situations that cannot always be scripted. To govern the technology, it must decide where authenticity ends and invention begins.

In other words, AI forces an organization to make tacit knowledge more explicit.

For an institution built on culture, that process may itself be valuable.

The Inn's project therefore represents a more nuanced vision of artificial intelligence than replacing labor with software. It asks whether an organization can use new technology to give people more opportunities to learn from the humans, stories, standards and ideas that made the organization exceptional.

That is an especially fitting experiment for The Inn at Little Washington.

Its history has never been about choosing between tradition and imagination. The more interesting pattern has been finding ways for each to strengthen the other.

Little Patrick—and the larger digital, archival and organizational work surrounding it—continues that pattern. The technology is new. The underlying ambition is not: preserve what matters, teach it generously, and keep finding new ways to make an experience unmistakably human.