Designing Vessa: A Hospitality-Informed Conversational System
Translating hospitality philosophy into operational AI behaviour within WhatsApp constraints.
Hospitality-to-Behavior Translation
From Hospitality Philosophy to Live Conversational System Design
Select a principle from the research section to see the mapped system translation and user-facing effect.
System Translation
- Character Bible defines role as curator, not search engine.
- Recommendation framing emphasizes warmth and discernment.
- Tone and pacing reinforce relationship-led service.
User-Facing Effect
- Service feels personal rather than transactional.
- Suggestions feel considered and context-aware.
- Trust builds through consistent conversational character.
Selected Principle
Hospitality is emotional, not transactional.
The project operationalized this principle through character definition, recommendation behavior, and relationship-led phrasing inside WhatsApp constraints.
Select each principle to view corresponding system translations and user-facing effects.
Executive Overview
Vessa is an AI personal restaurant concierge operating within WhatsApp, designed as a relationship-led service presence rather than a transactional assistant.
My role encompassed:
- • Authoring the foundational Character Bible
- • Designing Vessa's tonal system and interaction philosophy
- • Translating brand positioning into enforceable linguistic standards
- • Shaping conversational UX within WhatsApp constraints
- • Crafting tonally appropriate refusals and scope explanations
- • Creating a custom GPT to scale voice governance across the team
This work integrated hospitality philosophy, user experience design, brand strategy, engineering awareness, and production scalability into a cohesive conversational system.
Impact
Established a hospitality-led conversational architecture that preserved brand integrity under platform constraints, enabled consistent concierge-level tone at scale, and embedded voice governance into production workflows.
Research & Domain Immersion
Restaurants were not a domain I had previously worked within. To design with credibility, I immersed myself in hospitality literature and service philosophy.
A key influence was Unreasonable Hospitality by Will Guidara, which articulates how exceptional dining experiences are shaped not by efficiency alone, but by emotional attentiveness, anticipation, and care.
Several principles directly informed Vessa's design:
- • Hospitality is emotional, not transactional.
- • Care should feel effortless, not procedural.
- • Friction belongs to the service, not the guest.
- • Candour builds more trust than overconfidence.
- • Small gestures shape perception more than grand claims.
These principles became embedded in Vessa's philosophy:
- • Efficiency should feel like calm, not haste.
This research ensured that the digital concierge felt culturally aligned with premium dining hospitality.
The Character Bible: Philosophy Before Prompt
Before system prompts were written, I authored a foundational document: “Vessa – Character, Values, and Standards.”
This defined:
- • Identity and role (curator, not search engine)
- • Motivation and philosophy
- • Ethical boundaries
- • Taste standards
- • Decision logic under uncertainty
- • Micro-behavioural rhythms
- • Emotional pacing
To avoid generic AI tone, I anchored the character in tonal references:
- • Stanley Tucci — cultivated warmth and discernment
- • TOPJAW — contemporary insider authority
- • Tom Hiddleston — articulate composure
These references were distilled into operational attributes:
- • Calm confidence
- • Lightly opinionated guidance
- • Contained enthusiasm
- • Emotional discernment
- • Trust-first communication
The Character Bible acted as the single source of truth for all interactions.
Translating Character into Interaction Design
Character influenced not only tone, but structure.
Question Design
Questions were phrased to surface emotional context:
“What sort of evening are you picturing — cosy and quiet, or something a little livelier?”
This shifts the interaction from filtering to curation.
Recommendation Framing
Each recommendation balances:
- • Occasion and atmosphere
- • Food strengths
- • Practical reassurance
Examples of linguistic patterns:
- • “For a relaxed catch up, this feels like the better fit.”
- • “If you’d like something a touch smarter, I’d lean here.”
- • “Worth knowing it’s buzzy after eight.”
Word choice reinforces:
- • Discernment
- • Confidence
- • Emotional alignment
- • Concierge credibility
The experience feels considered rather than generated.
WhatsApp-Constrained UX Design
WhatsApp imposes:
- • Linear message flow
- • No persistent filter panels
- • High scroll friction
- • Limited visual hierarchy
Design decisions included:
- • Immediate value once onboarding is complete
- • Progressive refinement instead of interrogation
- • Structured restaurant blocks for thumb scanning
- • Maximum three recommendations per response
- • One clarifying question at a time
The result: emotional smoothness within platform constraints.
Tonally Appropriate Scope Boundaries
Booking integrations were architected at system level, but live availability and automated booking were not fully enabled at the time.
Vessa therefore:
- • Transparently communicated capability limits
- • Offered booking URLs or phone numbers
- • Translated system constraints into calm reassurance
For example:
“I can’t book that directly just yet, but I’ll share their details so you can secure it easily.”
This approach:
- • Protects trust
- • Maintains dignity
- • Keeps friction invisible
Hospitality principles shaped technical limitation handling.
The Living Lexicon: Scaling Governance
To prevent tone drift and reduce reliance on me, I built a custom GPT trained on:
- • The Character Bible
- • Approved examples
- • Linguistic constraints
- • Interaction patterns
This acted as:
- • A tone validation tool
- • A copy-generation assistant
- • An edge-case testing sandbox
- • A 24/7 accessible “living lexicon”
Strategic impact:
- • Accelerated production
- • Enabled asynchronous collaboration
- • Maintained voice consistency
- • Embedded governance into workflow
Character became infrastructure.
What This Demonstrates
Research-Led Design
Grounding conversational design in hospitality philosophy.
Brand-to-Behaviour Translation
Turning personality into operational linguistic standards.
Tonal Architecture
Designing phrasing, rhythm, and pacing as system components.
Platform-Aware UX
Delivering refinement and warmth within WhatsApp constraints.
Ethical Communication Design
Handling limitations with candour and care.
Production Enablement
Creating tooling to scale brand consistency.
Strategic Impact
This work demonstrates the ability to translate brand philosophy into structured conversational systems, align UX with engineering realities, and operationalise character governance at scale. Vessa became not just a chatbot, but a coherent service presence with enforceable behavioural standards and production-ready infrastructure.
Transferability
- AI Booking & Marketplace Agents Applying hospitality standards to automated service interfaces.
- Memory-Aware Conversational Products Using past context to deepen relationship-led service.
- Branded Assistant Workflows Scaling distinct voice and tone across large-scale deployments.