Work

Designer and builder · Scenra

Plan it for me. Hand it over.
Let me just drive.

Scenra is an iOS app built for road trips. It plans what to see on the way, where to refuel or charge, and what that costs. I design and review everything. Claude, an AI, writes the code.

2026 · Live on the App Store since 2 July 2026

Why I built it

I wanted to explore the UK and Europe by car.
No app planned that kind of drive.

In 2025 I planned a Lake District trip in Gemini and Claude. Every stop meant a question, a screenshot for my friends, and a copy and paste into Google Maps.

Ask

One stop at a time.

Where is the stop? Can I open it in Google Maps?

Screenshot

To show my friends.

Paste

Into Google Maps.

To know where to go next.

Repeat

For every stop on the trip.

The idea

The next trip was London to Edinburgh.
I built the app that plans the whole drive.

You say where you are driving, what you like to see and how many stops. Scenra plans the route, the stops, and where to refuel or charge.

Say where, and how many stops.
Say where, and how many stops
Pick what you like to see.
Pick what you like to see
Get the whole drive.
Get the whole drive

10 April 2026 · A test drive to Cheddar Gorge

The app sent us to a service station that did not exist.
The mistake was mine.

I was checking every scenic stop against Google Maps. Service stops skipped that check, so the AI's invented one went straight into the trip.

The hallucinated stop by Ai is a complete mistake of mine.. didn't pass service stops through the validation pipeline.

What I posted that evening, word for word.

My Instagram story from the road that day.
My Instagram story from the road that day

The decision

The AI can suggest. OpenStreetMap can suggest.
Nothing reaches you unless Google Maps confirms it.

Scenra draws on three sources. Each is good at one thing, so every stop from every source must pass the same check, service stops included.

Claude, the AI

Knows what is worth seeing.

It can also invent a place.

OpenStreetMap

Knows what is there: a viewpoint, a pass, a castle.

It cannot say if it is any good.

Google Maps

Proves the place exists, and how people rate it.

The final say on every stop.

My rules

Being real is not enough.
I wrote the rules for which stops are worth your time.

Real

On Google Maps, where it says it is.

Near

Within 10 km of your road.

Trusted

At least 10 reviews.

A 5.0 from one person ranks below a 4.4 from 653.

Scenic

A cathedral counts.

A campsite does not.

Different

One stop per landmark.

Not three views of it.

What changed

After the 10 km rule, no stop on my six test routes was far from the road.

My own test, run the same day before and after the change (1 August 2026): three routes in California, two in Britain and one in the Netherlands.

36 → 9 km The furthest stop from the road Across all 6 routes
4 → 0 routes with a stop over 10 km off the road Of 6 routes
75 → 92% of the stops asked for were delivered The map gave the AI more real places to pick from

The second worry

Will I run low, and what will it cost?
Scenra plans the trip around your actual car.

Add your car once. Scenra places the fuel or charging stops, shows the price at each one, and tells you what the whole trip will cost before you leave.

Your car and its limits.
Your car and its limits
The trip's cost, up front.
The trip's cost, up front
Where to charge, and the price.
Where to charge, and the price

The third worry

I used to copy every stop into Google Maps by hand.
Now one tap hands the whole route over.

Scenra plans the trip. It does not try to be your sat nav: Apple Maps, Google Maps and Waze already do the driving well.

When the plan is not quite right

The AI's plan is a first answer.
You can swap any stop, and the route redraws itself.

Swipe a stop to replace or remove it. Scenra offers places nearby with their ratings, then recalculates the drive.

On the road

I am the first user.
I test it the only way that counts: by driving it.

Since launch: Seven Sisters, North Wessex, and the Lake District with my mum and sister. Next: Cheddar Gorge again, Belfast, and Ireland in November.

Angel of the North.
Angel of the North
Lullingstone Castle.
Lullingstone Castle
Box Hill.
Box Hill
The white cliffs.
The white cliffs
And the day it failed on me.
And the day it failed on me

After launch

Once it was live, the numbers told me what to fix next.

97% of people planned on one day only, so I stopped selling a subscription and priced the single trip. Both results below are before and after, not controlled tests.

48 → 67% of new users plan a trip After I rebuilt the first launch to end in a trip
3 → 18% of new users pay One price per trip, £2.99 · An early signal, first 10 days
2 min 33 s from installing the app to a first trip Median

What I took from it

An AI can plan a trip in seconds.
The design work is making sure you never have to check it.

Back All work Moyinoluwa Olufisayo
Next Work with Mo moyinoluwafisayo@gmail.com

LinkedInResume