Tartendu Kumar AI Systems Engineer
00Index 01Work 02About 03Log 04Press 05Contact 06Résumé

Index / Work / MyPhoneAI — AI Call Receptionist

Mobile App Consumer AI Mobile

MyPhoneAI — AI Call Receptionist

Mobile app architecture and development

An Android app that gives you a second number answered by an AI receptionist. It picks up when you cannot, holds a conversation, takes reminders and forwards the calls that matter.

FlutterDartFirebaseRazorpayAndroid
No capture
01 Specification
Client
Zencia.ai
Industry
Consumer AI
Role
Mobile app architecture and development
Platform
Mobile
Stack
Flutter · Dart · Firebase · Razorpay · Android
Dated
2026-06-05
02 Key features
  • Twelve-step onboarding ending in a live test call before payment
  • AI answers, converses and forwards calls on your behalf
  • Call history with recording playback
  • Time-based and location-based reminders, restored after reboot
  • Knowledge sources — teach the assistant about you
  • In-app subscription and usage tracking
  • Thin client: no model inference on the handset
04 Problem → solution → outcome

① Challenge

Consumer users will not tolerate a setup flow that feels technical, but the product genuinely requires call forwarding, a purchased number, a configured AI personality and sensitive permissions. Get the onboarding wrong and the app is uninstalled before it ever answers a call.

② Solution

A twelve-step onboarding flow that introduces one decision at a time — name, personality, voice, capabilities, number — then a live test call before any payment is requested. The app is a thin client: no model inference runs on the phone, so the handset stays responsive and the assistant keeps working while the phone is asleep. Location-based reminders survive reboot through a restored geofence watcher.

③ Result

A shipped Android app covering onboarding, number purchase, call history with recording playback, reminders and in-app subscription. 49 Dart source files across 22 screens, with a companion dashboard for operations.

Need something
like this built?