From a Need
to a Real Product
A travel planning platform I designed, prototyped with AI, and built — because nothing like it existed.
MY ROLE
Sole Designer & Product Owner
STATUS
70% MVP Complete
BUILD METHOD
Figma + Cursor + Claude API
USER GROUPS
Travelers · Creators · Agents
01
ORIGIN
A Figma file was the whole plan
In 2025, I planned a two-week trip to London and Paris. Flights in one tab. Hotel confirmations buried in email. Restaurant saves across three apps. I did what designers do: I opened Figma and organized everything into frames I could actually use while traveling. It worked. But it shouldn't have to be a Figma file.
Planning our next trip — Portugal with two young children — I needed more. A live map showing saved restaurants near wherever we were standing. One tap to see the Airbnb address or the car rental confirmation code. A to-do list for visa applications and crib requests. A payment tracker so I knew what I'd actually paid before boarding. I needed this to be a real product.
"I couldn't find something that would compile all of this into one place — so I built it."
02
THE PROBLEM
Travel planning is 6 tools duct-taped together
If you travel like me, you stitch together Google Maps, email, spreadsheets, notes apps, calendars, and PDFs to plan a single trip. None of these talk to each other. None know you're planning a trip. None can tell you that you haven't booked transport between two cities, or that you're double-booked on a Thursday afternoon.
After mapping the competitive landscape, two things became clear: no product serves the full end-to-end planning experience in a simple way, and nobody has built the creator layer — a marketplace where travel creators sell usable itineraries that convert buyers into subscribers. That's the white space.
03
THE SOLUTION
One product, three distinct experiences
The core architectural decision: a shared trip infrastructure — map, calendar, documents, AI — with distinct surface layers for each user group. Same foundation. Three different entry points and mental models.
GROUP 01
Travelers
Individuals, couples, and travel companions. Seat-based subscription with all core planning features.
SOLO $9 · DUO $15 · CREW $22 /mo
GROUP 02
Creators
Travel bloggers who build and sell itineraries. Buyers get read-only access — editing requires a Group 1 subscription, creating a built-in conversion funnel.
CREATOR PLAN + 20% MARKETPLACE FEE
GROUP 03
Agents
Travel agencies who generate branded, client-ready itineraries. White-label branding, tiered itinerary credits, team seats.
TIERED — $49 / $99 / $199 /mo
04
FEATURES
Every feature traces back to a real frustration
Interactive Trip Map
Saved locations pinned and filterable by day. The primary use case: standing somewhere unfamiliar, opening the map, seeing what you've saved nearby.
Calendar Itinerary Builder
Day-by-day event blocking with AI logistics detection running against the calendar in real time — flagging gaps, missing transport, and timing conflicts before departure.
Document Vault
One-tap access to passport photos, visa confirmations, hotel codes, car rental PDFs. Trip-scoped and shared across all seats in a subscription.
Payment Tracker
Paid / Pending / Due status tags across all trip expenses. Know what you've actually paid before you board — no spreadsheet required.
Trip To-Do List
A trip-scoped checklist for pre-travel logistics: visas, crib requests, travel insurance, passport photos. Everything visible is relevant to the current trip.
Post-Trip Dashboard
A visual recap of completed trips — places visited, total spend, countries covered. The social and emotional payoff that drives word-of-mouth.
and it took a bit of back and forth to get these all set up...
05
DESIGN CHALLENGES
The decisions that were actually hard
The AI Logistics Layer
The challenge isn't building AI detection — it's designing the surface. Too aggressive and it becomes noise. Too passive and it fails the core use case. The product needed three alert levels: inline calendar flags for timing issues, a logistics panel for gaps requiring action, and a pre-departure summary that consolidates everything into one actionable checklist.
Resolution
Alert copy is plain-language and specific ("You haven't booked transport from London to Paris — Eurostar is usually 2h 15m"), not AI-hedged. Color hierarchy: amber for warnings, red for blockers. The AI runs against actual trip context — destinations, dates, booking status — using the Claude API to reason about real-world constraints.
Google Maps API Integration
Maps API is built for navigation, not planning. Default styling, pin behavior, and clustering are tuned for the wrong context. Making it feel native required rebuilding how pins communicate — type, day, and status — while staying readable at any zoom level.
Resolution
Pin color maps to trip day, not category. Category communicates through icon shape. The sidebar stays visible when a pin is selected — split view, not replacement — because losing map position when viewing details was the most common friction point in competing products.
Three Groups, One Product
The risk with multi-persona products is designing for a lowest common denominator that serves no one well. The solution was treating shared infrastructure as invisible — same map, calendar, and documents under the hood — and designing three distinct first-run experiences that surface only what's relevant to each group.
Resolution
Onboarding acts as a group selector. Creator and agent features exist in the product but are never surfaced during the traveler flow. Users only encounter capabilities relevant to their context until they choose to expand.
04
AI IN THE PROCESS
Designing and building with AI throughout
This project used AI at two distinct levels: as a build tool to close the designer-developer gap, and as a product feature powering the logistics engine. Both changed how I work.
DESIGN IN FIGMA
Flows, components, interaction states
PLAN WITH CLAUDE
Scope features, define details, generate dev-ready spec docs
PROMPT IN CURSOR
Use output from Claude to prompt Cursor to create
LIVE PROTOTYPE
Real data, real interactions, real constraints
ITERATE
Figma, Claude, Cursor — on repeat
What this changes about the designer's role
Building the prototype myself — rather than writing specs for a developer — meant I could make and test a decision in the same hour. Technical constraints became design inputs, not post-hoc blockers. This is now a core part of how I practice design.
04
ROADMAP
What happens next
Now
Complete Group 1 MVP, beta launch
Finish AI logistics engine, Stripe billing, and post-trip dashboard. Recruit 50 beta users. Target ProductHunt launch and $500+ MRR.
3–6 months
Creator marketplace + iOS beta
Build Group 2 — publish flow, buyer paywall, Stripe Connect payouts. Seed marketplace with 20+ itineraries. iOS beta on TestFlight.
6–12 months
Agency platform + iOS public
White-label branding, tiered itinerary credits, agency team seats. iOS public release. Target $8k+ total MRR across all three groups.
12–18 months
$20k MRR, hire or raise
All three groups active. Evaluate bootstrapped growth (one contract developer) vs. pre-seed raise based on retention and MRR trajectory.
04
REFLECTION
What I learned
On being your own user
The Figma trip file was the most useful design artifact I made on this project — not because it was good design, but because it was honest. Every workaround was a product requirement. Designing from lived frustration produces a different kind of conviction than designing from personas.
On multi-group products
The instinct is to design each group's experience in isolation, then find the overlap. The better approach is the reverse: start with shared infrastructure and work outward. What three different users need from the same map is more similar than it first appears — the differences are almost entirely in surrounding context, not core interaction.
On AI in the product
The AI logistics layer was the most technically complex feature and the one I was most uncertain about until it worked. The hard problem wasn't the AI — it was the surface. Impressive capability that appears at the wrong time, in the wrong tone, or with too much friction is worse than not having it at all. The work was mostly copywriting and timing.
On building with AI
Using Cursor to build the prototype changed how I understand the designer's role. The closest feedback loop to great design is the one that runs through implementation. When I can make and test a decision in the same hour, the design gets sharper faster. This is how I work now.
Want to try this?
Get notified when I release