← All work
Social · Full-stack

Balsam

An app where people talk to trained listeners by chat, audio or video call. We built all three parts of it for a client in Dubai.

Client
Balsam · Dubai, UAE
Platform
Mobile apps · Android & iOS, plus admin panel
What we did
User app, listener app, admin panel and backend
Timeline
May 2026 – ongoing
Balsam screen 1
Balsam screen 2
Balsam screen 3
Balsam screen 4
3 apps
user, listener and admin
Chat, audio, video
every way to talk to a listener
Arabic & English
with full right-to-left support

What happened on this project.

What it is

An app where people talk to trained listeners when they need someone to hear them out. Users browse listeners by category and language, then chat, call or video call them, paying with coins. Listeners apply through their own app and are verified before they go live.

What we did

We built the user app, the listener app, the admin panel and the backend behind them. That covers chat with voice notes and media, audio and video calls, random matching, coin purchases through Stripe, gifts, reviews and listener verification.

What brokeCalls didn’t come through when the app was in the background.

If the app was closed or the phone was locked, a user could call a listener and it would never ring. For an app built around calls, that was the core of the product failing.


What we did about itWe built native background calling with a wake-up.

The phone now wakes up and rings like a normal call, even when the app is closed, so listeners can pick up from the lock screen.

Balsam screen 1
Balsam screen 2

From me, and from the client.

Shahzad Umar BaigFounder & principal engineer

The calling was built on packages that weren’t right for the job, so I replaced them with a native implementation. It took two weeks because it was major work, but once it shipped the calling bugs were gone, and Dr. Ali was happy.

Dr. AliFounder · Balsam · Dubai, UAE

“I truly hope our experience with Balsam will be successful, because I see this project as the cornerstone of a much larger long-term business vision. I truly appreciate your efforts and hope we continue building a successful long-term partnership.”

More screens from the app.

Send a gift
Send a gift
After a call
After a call
Call history
Call history
All listeners
All listeners
Built withFlutterNode.jsMongoDBSocket.IOZEGOCLOUDCallKitFirebase Cloud MessagingStripeAWS S3Next.js

Next case study

← Back to all work
AI · LLM pipeline

YDigest

500K → 10K API calls

An app that turns YouTube videos into transcripts and summaries. It came to us as a vibe-coded prototype, and we rebuilt the data layer so it could handle many more users. It was making 500,000 API calls. After caching and fixing memory leaks, it makes 10,000.

Read the case study →
YDigest screen 1
YDigest screen 2
YDigest screen 3