06 / TOO USEFUL TO KEEP TO OURSELVES
Made for us.
Shared with our friends.
It started with the two of us and eventually became something we shared with friends. That shift is part of the fun: a personal workaround can become something worth passing on. A small idea gets a life beyond the people it started with.
The best part? We didn't have to stop at “someone should make that.”
POSTCARD FROM A SIDE PROJECT
Small friction.
A conversation.
A thing that actually exists.
Build
Designed & directed by me. Coded with AI
Our trips needed
adult
supervision.
So I made them a little home on the internet.
A shared spreadsheet became a travel app. Built through conversations with AI, improved through our own trips, and shared with friends.
vibe coded.
life tested.
LESS “WHERE'S THE TICKET?” ✳ MORE “WHEN DO WE LEAVE?” ✳ LESS “WHERE'S THE TICKET?” ✳ MORE “WHEN DO WE LEAVE?” ✳
01 / THE VERY RELATABLE BEGINNING
Our travel was spread across too many places.
Some tickets were booked directly with IndiGo, Thai Airways, AirAsia or Air India. Others came through Indian Railways or a booking platform. Some journeys were personal; some were together. Remembering who had booked what—and where the confirmation lived—became unnecessarily confusing.
Because we were long-distance for parts of the year and frequently travelled home, the work started much earlier than the booking. We had tentative dates to align, journeys still to book and Rishi’s days in Thailand to calculate.
The information lived across WhatsApp, email confirmations, calendars and memory. We used excel to plan stuff sometimes were always confused and every “when are you coming?” started another round of searching.
How might we bring confirmed bookings, unfinished plans and residency-day calculations into one shared, dependable place?
02/ THE “CAN WE JUST…” MOMENT
What if I could just ask for it?
I started building through conversations with ChatGPT, often from my
phone. Small requests. A working version. A real trip to try it on. Then
another request.
No hand-written code. I shaped the features, supplied layouts where
needed, checked the behaviour and nudged the design. AI turned those
conversations into implementation.
● ● ●
A VERY SHORT VISUALISATION
A
Can we put our trips in one place, show who's travelling, and keep plans that aren't booked yet?
A first version to try...
A
Okay, useful. But this trip is for both of us. And that date is Bangkok time. Also… can we change that
icon?
build → use → spot friction → ask again
MY PART
The problem, priorities, layouts, rules and “that doesn't
feel right.”
AI'S PART
The code, implementation and help interpreting ticket
information.
THE REAL TEST
Could we actually use it to coordinate our next journey?
03 / IT DEVELOPED
And then came the “what ifs…”
We built almost entirely through the chat interface on our phones. We gave it small requests, used the result, then returned with the next issue—an icon that felt wrong, a label that was unclear or a planning state that was missing. Early on, usefulness mattered more than a perfect interface. Visual consistency, responsiveness and accessibility were strengthened through later refinement passes.
Win
No more
hunting
Version 01 · Visibility
“Where's that ticket again?”
The Excel sheet proved that a shared source of truth reduced searching. The first secure site kept that structure, but made the next departure, traveller and ticket details immediately visible.
—
One protected shared timeline
—
Rishi, Aanchal or both
—
Booking source and ticket in context with quick snapview
Version 02 · Planning
“But we haven't booked it yet.”
Real travel exists before the booking confirmation. Dashed “Yet to be booked” states made intention visible without pretending that details were final.
—
Planned and booked coexist
—
Visual key of what has to be done
—
Upload ticket when ready
Win
Intentions
stay visible
Win
Bigger
picture
Version 03 · Orientation
“What does the whole month look like?”
A calendar made gaps and overlaps easier to spot. A dedicated Thailand-days view then used confirmed and planned journeys to make residency planning part of the same travel system.
—
Weekend and today emphasis
—
Confirmed and planned-day counts
—
Consistent arrival/departure rules
Version 04 · Intelligence
“Surely I don't have to type all this?”
GPT helped interpret cities, airport codes and ticket details.
Product rules connected related journey legs into Smart Trip
Hubs.
—
City and airport-code parsing
—
Related tickets grouped automatically
—
Suggestions stay attached to the journey
04 / THE GLOW-UP WAS MOSTLY MENTAL
Same trips.
Less brain admin.
Before
Four places to reconstruct one journey.
What's next, who's going, what still needs booking.
After
05 / VIBE CODING STILL NEEDS A BRAIN
Fast to build.
Still things to think through.
AI made it easier to get a version in front of us. Using it revealed the
decisions that actually mattered.
IDEA
Useful first. Prettier as we went.
The early priority was seeing the next journey and finding its details. I refined icons,
copy, layouts and responsiveness as we used it.
Ownership
AI interpretation ≠ product rules.
Reading “BKK” is one task. Deciding which journeys belong together is another. Parsing used GPT; grouping and travel states needed consistent rules.
Time
A plan and a booking aren't the same.
Both deserve visibility. Clear states let us plan ahead without making an unbooked journey look confirmed.
usecase
Shared doesn't mean identical
Traveller ownership stayed visible, with filters for Rishi, Aanchal or both. One shared source could still answer personal questions.
A useful app. A fun experiment. And a very good excuse to keep asking, “what if?”
Win
less repetitive
entry
Outcome
A personal tool that reduced coordination to a glance.
Shared clarity
Both people can see what is
confirmed, what is pending and
who is travelling.
Less re- checking
Dates, times, tickets and trip
context no longer need to be
reconstructed from chats.
A system that grows
New edge cases fit into a small
set of rules instead of creating a
new interface every time.
© 2026
resume (PDF) →
Reminder to you & me : Let’s go touch some grass.
built with love, brain and AI in the slowly burning world…
Explore more work
Next project ↗
Go Top