Functional, until the network disappeared.
The first version worked online but failed the exact traveller who needed SafePak most: someone at an airport without a SIM, working data, or a reliable connection.
DiscardedSafePak started with a simple gap: visitors arrive with a phone full of internet, but without the small pieces of local knowledge that make a place feel navigable. A fair taxi fare. A working app. A number that actually answers.
Enough local context to know what is normal before agreeing to a ride.
Visitors do not automatically know the things a local already knows.
A foreign YouTuber, filming his first hours in Pakistan, needed to get from the airport to his hotel. He did not know which apps operated locally, what a fair street taxi fare looked like, or where to verify either answer.
He took a street taxi, paid roughly double the fair fare, in a car barely holding together, and said so on camera to an audience deciding whether to visit.
That was not a story about a malicious country or a bad driver. It was a story about missing context. SafePak exists to make that context easier to find before a visitor learns it the expensive way.
The first versions were not failures because the code was poor. They failed because they optimised for the wrong constraint.
The first version worked online but failed the exact traveller who needed SafePak most: someone at an airport without a SIM, working data, or a reliable connection.
DiscardedThe second version added serious infrastructure, admin two-factor, cache invalidation, offline architecture, before the information itself was mature enough to deserve it.
OverbuiltThe version you are reading runs as static files with the emergency page cached on your device, so it opens in airplane mode. That took three attempts to understand.
CurrentThe product only becomes useful when design, cost and research are treated as product infrastructure rather than polish around the edges.
For a visitor deciding whether to trust unfamiliar information, clarity and presentation are part of the safety model. It took someone telling us plainly before we rebuilt it.
Every service behind this runs on a free tier, chosen deliberately so the product survives whether or not anyone ever writes us a cheque.
Service listings, emergency numbers and prices require current sources, calls, visits and reading the fine print, over and over. There is no database to buy.
An honesty rule means very little if a product only talks about the mistakes it caught before anyone saw them.
Our first prototype showed a confident message saying your alert had been sent to the police. It sent nothing. It also labelled guides as background-checked when nobody had checked anything.
Both were deleted rather than launched, and that decision became the rule the whole product now runs on.
For a while this site told arriving visitors to install two ride-hailing apps that no longer operate in Pakistan. Uber left in 2024 and Careem ended its rides in 2025, and we were still recommending them from memory.
We now verify service listings against current sources rather than assumption.
Small team, deliberately narrow scope, and a product that has to earn trust through the quality of its information and the usefulness of its experience.
NUST
NUST
NUST
Air University
In three years, SafePak intends to be the single trusted bridge between foreign visitors and Pakistan’s services, its institutions, and its people.
SafePak vision
Open SafePak to see the current visitor experience, or contact the team if you represent a tourism body, mission, university or potential research partner.