MyTravelBuddy.ai · In progress
MyTravelBuddy.ai — a travel companion that actually travels with you
An AI-powered companion designed around the journey after planning: where the traveler is, what is happening next, what changed, who is traveling, and what would be useful right now.
Role: Founder / Builder
- Context-aware AI
- Offline-first mobile
- Travel lifecycle
Origin
The story
MyTravelBuddy.ai began with a simple idea: planning the trip is only the beginning.
I have used and enjoyed travel products such as Wanderlog. They are excellent at organizing flights, hotels, reservations, activities, and itineraries. But while traveling with my family, I kept encountering a different need. Once the trip started, I did not need another itinerary editor. I needed a partner.
I wanted something that knew where we were going, understood the itinerary, remembered who was traveling, knew our preferences and what we had already done, and could help with the constantly changing questions that happen during a real trip.
Where should we eat near where we are now? What should we do because the weather changed? How much time do we have before our next reservation? Which entrance should we use? When should we leave? What have we already seen? And sometimes: something went wrong—what do we do now?
MyTravelBuddy grew out of that experience. The goal is not to build another trip planner. It is to build the companion I wanted traveling alongside my family.
The central principle is simple: the traveler should be the center of the system, not the itinerary.
Journey
A companion built around real travel
From trip planner to travel companion
Traditional travel products are primarily built around the itinerary. MyTravelBuddy is built around the traveler.
The itinerary remains important, but it becomes one part of a larger, living context: who is traveling, where they are, where they are going next, what they enjoy, what they have already done, reservations, weather, transportation, family routines, and changes that happened along the way.
The Buddy maintains that context throughout the journey instead of treating every question as a new search. The experience should gradually feel less like querying travel software and more like asking: “Buddy, what should we do now?”
Built from real travel
The product is being designed around my family’s actual trips rather than hypothetical travel workflows. A recent Alaska and Vancouver journey became a particularly useful test case.
The questions changed constantly with our location, schedule, weather, and circumstances. None belonged neatly inside an itinerary, but together they represented the actual experience of traveling.
That became a core design principle: travel is dynamic. The product needs to understand the journey, not merely store the plan.
- What should we see at our next cruise stop?
- Why is the ship waiting outside a glacier?
- Which Grouse Mountain gondola should we take?
- What can we do indoors because it started raining?
- Where can we eat and shop between Stanley Park and Richmond?
- Is there something interesting nearby for the kids?
- What happens when a phone gets lost during the trip?
- How do we arrange recovery and shipping after it is found?
The travel lifecycle
I designed MyTravelBuddy around five stages. The Buddy remains present as a vague idea becomes a plan, a plan becomes a real journey, and that journey becomes useful memory for the next one.
- Dream — turn loose ideas such as “somewhere warm” or “a week with the kids” into possibilities while learning what the traveler enjoys.
- Plan — organize destinations, flights, hotels, cruises, activities, restaurants, transportation, reservations, and daily itineraries.
- Travel — shift into a context-aware Today experience that surfaces what matters now rather than making people navigate the plan.
- Remember — gently connect photos, places, activities, notes, favorite moments, and unexpected discoveries into a travel journal.
- Grow — carry useful preferences and routines across trips so future travel becomes increasingly personal.
Context-aware assistance
Travel recommendations are rarely universally correct. A great recommendation at 10:00 AM may be completely wrong at 4:30 PM when dinner is booked across town at 6:00.
MyTravelBuddy combines situational signals before deciding what information is useful. Instead of requiring the traveler to assemble the context manually, the Buddy grounds the conversation in the actual journey.
- Current location and local time
- Weather and transportation conditions
- Upcoming reservations and available time
- Who is traveling and what they prefer
- What has already been visited
- What is nearby and realistically reachable
Mobile is part of the architecture
The web application is ideal for dreaming, planning, itinerary management, and trip administration. Active travel introduces requirements that a browser cannot reliably solve.
The mobile experience is therefore being designed around React Native and Expo, with native capabilities where the journey needs them. The web application helps create the trip; the mobile application becomes the companion that actually comes along.
- Background location awareness and geofenced itinerary events
- Push notifications and context-aware traveler alerts
- Local device storage and offline operation
- Synchronization when connectivity returns
- Camera, photo, and device-native sharing
- Reliable trip context while the app is backgrounded
Offline-first travel
Travel apps are often needed most precisely where connectivity is worst: airports, cruise ships, foreign cellular networks, remote destinations, underground transportation, and mountains.
Essential trip information should be cached locally. Actions taken offline should be recorded on the device and synchronized when connectivity returns. A travel companion should not become useless because the traveler lost signal.
- Today’s itinerary and upcoming reservations
- Addresses, confirmation details, and notes
- Relevant traveler and trip context
- Local mutations with background synchronization
- Explicit conflict handling across devices
Travelers, observers, and arrival intelligence
Trips involve more people than the travelers themselves. With permission, observers can follow selected aspects of a journey without requiring constant manual updates.
This is not intended as continuous surveillance. The idea is to translate location into meaningful travel events. A coordinate alone says very little; location plus itinerary plus time creates context.
Entering the area around an airport that corresponds with today’s scheduled arrival can suggest that the traveler has landed. The same principle can apply to hotels, cruise terminals, attractions, restaurants, tours, and activities—supporting both traveler assistance and carefully scoped observer notifications.
Memory has to earn its place
Most travel applications forget you when the trip ends. MyTravelBuddy is intentionally designed not to.
The Buddy can learn useful preferences such as scenic drives over highways, a grocery stop after checking into a rental, museums in the morning, favorite activities, travel pace, and places loved or avoided.
The goal is not to collect personal information simply because the system can. Memory must earn its place by making future travel meaningfully better.
AI is the interface, not an attached feature
MyTravelBuddy is not an itinerary application with a chatbot attached. Conversation is part of the interaction model.
Travelers should be able to ask what is worth doing nearby, whether there is enough time before dinner, what should change because it is raining, where they ate yesterday, or what needs to happen tomorrow morning.
The AI layer combines natural conversation with structured trip data, persistent memory, location, time, and external tools so answers remain grounded in the real journey.
The product philosophy
Flights change. Weather changes. Plans change. People get tired. Kids get hungry. Phones get lost. Unexpected places become the highlight of the trip. Travel software should adapt to that reality.
The best travel companion is not the one that created the perfect itinerary three months ago. It is the one that is still useful when you are standing somewhere you have never been and asking: “Okay, Buddy. What now?”
Context
The problem and my role
Problem
Travel products are excellent at organizing an itinerary, but the real trip is dynamic. Weather changes, reservations move, people get tired, connectivity disappears, and useful advice depends on location, time, companions, and what has already happened. Travelers need continuity and situational help—not another itinerary editor.
Role
Designing and building the product end-to-end: lifecycle and companion UX, structured trip data, conversational reasoning, persistent memory, mobile and offline architecture, location-aware events, observer privacy, integrations, and the live product narrative at mytravelbuddy.ai.
Constraints
- Retain useful long-term context without making memory intrusive
- Combine probabilistic AI reasoning with deterministic itinerary and reservation data
- Use background location without excessive battery consumption
- Translate raw geographic signals into meaningful, explainable travel events
- Keep essential trip information available when connectivity is unreliable
- Synchronize offline changes and resolve conflicts across travelers and devices
- Support observers without turning location sharing into continuous surveillance
- Deep-link into a fragmented ecosystem of airlines, hotels, maps, and activity providers
- Know when proactive assistance is valuable—and when the system should remain quiet
System
Architecture
System map
The travel lifecycle
The Buddy carries useful context forward as an idea becomes a trip, a trip becomes a lived experience, and that experience improves the next journey.
- 01
Dream
Explore possibilities and learn what kinds of experiences the traveler values
- 02
Plan
Organize destinations, reservations, transportation, activities, and daily itineraries
- 03
Travel
Surface what matters now using live context and the structured plan
- 04
Remember
Connect photos, places, notes, and favorite moments into the story of the trip
- 05
Grow
Carry useful preferences, routines, and prior experiences into future travel
System map
From trip signals to useful assistance
Deterministic trip data and live situational signals ground AI reasoning before the system decides whether to answer, alert, remember, or remain quiet.
- 01
Structured trip
Travelers, itinerary, reservations, destinations, and transportation
- 02
Live context
Location, local time, weather, movement, connectivity, and nearby places
- 03
Context engine
Determines what is happening now and which constraints matter
- 04
Buddy reasoning
Combines conversation, tool calls, structured outputs, and persistent memory
- 05
Traveler assistance
Answers, recommendations, navigation, reminders, and carefully timed proactive help
- 06
Events and memory
Permissioned observer updates, journal moments, and useful context for future trips
Execution
What I built
- Defined a five-stage lifecycle—Dream, Plan, Travel, Remember, Grow—with continuity across the entire journey.
- Designed a Travel Today experience that combines schedule, location, time, weather, transportation, reservations, companions, and nearby places.
- Separated the planning-oriented Next.js web experience from a React Native / Expo architecture for active travel.
- Designed local caching, offline mutations, background synchronization, and conflict handling for unreliable connectivity.
- Modeled trips, travelers, destinations, itineraries, flights, hotels, cruises, activities, reservations, and transportation as structured context for AI reasoning.
- Designed geofenced itinerary events and arrival detection that translate coordinates into meaningful journey context.
- Introduced permissioned observer experiences around selected travel events rather than continuous raw-location sharing.
- Made persistent memory a product capability for preferences, routines, prior experiences, and successful itinerary patterns.
- Kept discovery provider-neutral through maps, weather, search, and deep links to airlines, hotels, cruise lines, and activity providers.
- Included lightweight journal capture for photos, places, notes, favorite moments, and unexpected discoveries.
Tradeoffs
Technical decisions
Center the traveler, not the itinerary
- Decision
- Treat the itinerary as one input to a broader traveler context that persists before, during, and after the trip.
- Why
- Real travel questions depend on people, place, time, preferences, history, weather, and change—not just what was scheduled.
- Tradeoff
- A richer context model is more useful, but requires clear ownership, privacy boundaries, and disciplined decisions about what deserves to persist.
Web for planning, native mobile for traveling
- Decision
- Use Next.js for planning and administration while designing the active-trip companion around React Native and Expo.
- Why
- Background location, geofencing, push notifications, local storage, camera access, and reliable background behavior are architectural requirements during travel.
- Tradeoff
- Two clients increase product and synchronization complexity, but forcing active-travel requirements into a browser would create a less reliable companion.
Offline is a data-model concern
- Decision
- Design essential trip data, local mutations, synchronization, and conflict resolution as first-class product behavior.
- Why
- Connectivity failures are normal during travel, so offline behavior cannot be postponed as a presentation-layer enhancement.
- Tradeoff
- Offline writes and shared trips require explicit consistency rules and more careful testing than an always-online client.
Meaningful events instead of raw coordinates
- Decision
- Combine geofences with itinerary semantics and time before creating arrival events or observer notifications.
- Why
- Location plus itinerary plus time can explain that a traveler likely arrived at a scheduled place; a coordinate by itself cannot.
- Tradeoff
- Semantic arrival detection is more useful and privacy-preserving, but must tolerate noisy location signals and avoid overstating certainty.
Ground AI in structured journey state
- Decision
- Use conversation as the interface while grounding answers in deterministic trip data, current context, tools, and persistent memory.
- Why
- The traveler should not need to reconstruct schedule, location, companions, and constraints in every prompt.
- Tradeoff
- The system must distinguish known facts, tool results, memories, and model inference so helpful language does not hide uncertainty.
Proactivity must earn attention
- Decision
- Make the Buddy context-aware enough to assist proactively, but require relevance, timing, and confidence before interrupting.
- Why
- A companion that stays silent during a useful moment fails; one that constantly interrupts becomes another source of travel stress.
- Tradeoff
- The right threshold varies by traveler and situation, so controls and learned preferences must remain understandable.
Value
Product impact
User impact
Helps travelers decide what to do now, adapt when plans change, retain essential information without connectivity, and carry useful preferences and memories into future trips.
Business impact
Creates a differentiated product position beyond itinerary management by making context, continuity, and active-trip assistance the durable value.
Results
Outcomes
- A product architecture grounded in real family travel rather than an abstract itinerary workflow.
- A clear differentiation from planning-first tools: the plan is an input; the traveler and the changing journey are the product center.
- A coherent web, native mobile, location, offline, memory, AI, and notification roadmap for an active-travel companion.
- A live product presence and early feedback loop at mytravelbuddy.ai.
Reflection
Lessons learned
- “Travel is dynamic. The product needs to understand the journey, not merely store the plan.”
- “Location becomes useful when combined with itinerary semantics and time; raw coordinates are not meaningful context on their own.”
- “Memory should not exist because it can—it has to earn its place by making future travel meaningfully better.”
- “A trustworthy companion must know both when to help and when to remain quiet.”
Tools
Stack
Next case study
DocuSend.io — making document workflows feel like software →