Finding the Nearest ATM in Hawassa, A Telegram Bot Powered by Gebeta Maps

Nov 9, 2025•3 min read
Finding the Nearest ATM in Hawassa, A Telegram Bot Powered by Gebeta Maps
Finding the Nearest ATM in Hawassa, A Telegram Bot Powered by Gebeta Maps

Hawassa

Finding the Nearest ATM in Hawassa, A Telegram Bot Powered by Gebeta Maps

Picture this:

You just arrived in Hawassa for work. It’s late, stores are closing, and your guest house doesn’t take card payments. You reach for your wallet, only to realize you’ve got almost no cash left. You don’t know the city. You don’t know the directions. You just need an ATM. Fast. That’s exactly the problem this project tries to solve. I built a Hawassa ATM Locator Bot, a Telegram bot that helps anyone quickly find the 5 nearest ATMs to their current location or by typing the name of the neighborhood.

How Did I Build This?

One Sunday afternoon, I drove across Hawassa collecting ATM locations—street by street, sub-city by sub-city. A friend of mine joined and finished the other half.

  • We tried to cover every corner
  • We documented each location carefully
  • We ended up with 99 ATM spots in Hawassa

The data lives in a CSV file:

id,name,locationa,locationb
1,CBE ATM - Shafeta,7.029828,38.488396
2,Sidama Bank ATM - Tabor Shafeta,7.029735,38.488440
...

What the Bot Does

The bot is straight forward. It lets you share your Telegram location OR type area name.

If location is sent:

  • Haversine formula finds the closest 10 ATMs
  • Gebeta Maps ONM API finds real road distance between you the machines
  • Bot returns the closest 5 ATMs

If name is sent:

  • Fuzzy search matches spelling
  • Bot returns nearest 5 matches

You might be thinking, why haversine? Haversine calculates the shortest straight-line distance between two points by accounting for the curvature of the Earth. But it doesn’t account for mountains, valleys and roads. That’s where Gebeta maps one-to-many API becomes handy.

Why ONM Matters

Haversine gives straight-line distance, while Gebeta ONM calculates real driving and walking roads.

In our case, Hawassa has two hills (Tabor/Dume & Alamura). Straight-line might say “2 minutes walk.” But the actual road might take more than 20 minutes around a mountain.

Gebeta ONM only takes 10 destinations. So, the bot uses haversine for shortlisting 10 ATMs relatively close to you and it uses ONM for real accuracy, to sort them as the top 5.

MethodResult TypeAccurate in Real Life?
HaversineStraight line❌ No (mountains, blocked roads, detours)
ONMRoad distance✅ Yes (humans don’t fly… yet)

Tech Stack

  • Node.js
  • grammY.js
  • Gebeta Maps ONM API
  • CSV dataset of ATMs

Final Output Example

This fun project is just a beginning. You can extend the same codebase to map clinics, restaurants, fuel stations, pharmacies or any other essential places people might need. You might even expand the idea to other cities. The foundation is ready... now it’s your turn to build on it. Happy hacking!

Links