hiRobin

The short version

Problem. Signup-to-activation sat at ~30–40%, and engineering tracked only one of the three ways OTP verification was failing.

Action. Pulled funnel data and session recordings to expose the two invisible failures, pushed the team to test a provider switch, then rebuilt onboarding end to end for iOS and Android.

Result. Activation rose to ~80%+ and D7 retention peaked at 74%, as the app grew to roughly 1 million downloads and 160,000 to 200,000+ active users.

The impact

• ~30–40% → ~80%+ signup-to-activation conversion
• ~1M downloads within 6 to 9 months
• 160K–200K+ active users
• 74% peak D7 retention

Two hiRobin screens on Android phones: the settings screen with incoming and outgoing call controls, and the Add Instructions chat where Robin summarises how to handle a type of call.

The problem hiRobin solves

India has a specific, real version of this problem: a high volume of spam and scam calls, a high volume of calls people genuinely need to make or receive, and very little time to handle either, especially for working professionals and business owners running on packed schedules. A personal assistant would solve this, screen the noise, make the calls you don't have time to make. Almost nobody can actually afford one.

hiRobin sits on your phone through call forwarding and acts as that assistant. Inbound calls get forwarded to Robin, who picks up, screens the caller, and understands what the call is actually about. If it's something Robin can resolve directly, a delivery update, for example, Robin handles it and reports back. If not, Robin transfers the call to you, with the context already gathered. Outbound works the same way in reverse: tell Robin to book a reservation, call a clinic to set an appointment, inquire about pricing or stock availability, or convey a message, and Robin makes the call.

Step through the product

1. Robin answers, you stay in the loop hiRobin sits on your phone through call forwarding. Robin picks up, screens the caller and works out what the call is about, and the home screen shows what it is handling and how many calls it has taken.
2. Context, not just a missed call If Robin can resolve a call itself, like a delivery update, it does and reports back. If not, it transfers the call to you with the context already gathered. Every call keeps a summary, the recording and a one-tap callback with Robin.
3. Hand a task to Robin Outbound works in reverse. Tell Robin to book a reservation, set a clinic appointment, ask about pricing or stock, or pass on a message, and it makes the call, checking the details with you in chat first.
4. Not another app: the dialer itself The direction being pursued is to make hiRobin the default Android dialer. Robin stops being a separate destination and becomes how the phone already works: make calls as usual, or hand any call or task to Robin from the same screen.
5. You decide how much Robin does Settings keep control with the user: silence unknown numbers, let Robin handle calls you decline or miss from contacts, pause Robin for a set time, and show a caller ID banner so you can see who is calling before you answer.

The diagnosis

Growth was stalling at the very first step: signup-to-activation conversion was badly bottlenecked, and engineering's own failure tracking couldn't explain why. Their tracking only counted one failure mode, wrong OTP code entered. Pulled funnel data and session recordings directly and found two more, both completely invisible to the existing tracking.

A smaller but real friction point layered on top: no auto-copy, users had to leave the app, open Messages, and return mid-flow.

The pushback

Presenting "your system has a failure mode you can't see" to a team confident in its own system didn't land in one meeting. Kept returning with sharper, more specific evidence, never-sent vs. delayed vs. wrong-entry, exact breakdowns, not a vague claim, until the team agreed to just test a provider switch. That test broke the deadlock: OTP failures dropped sharply once the switch happened, and the team could finally see what the tracking gap had been hiding. From there, onboarding was rebuilt end to end, platform-specific flows for iOS and Android, the mechanism behind the growth numbers above.

Beyond onboarding

Owned monetization design end to end ahead of Series A: pricing, packaging, paywall conversion testing, tier packaging, and feature-level gating across the full active user base. Sat on the final hiring panel, after two founder rounds, for the company's founding Product Manager and Social Media Marketing Manager. Partnered directly with marketing to structure the product's brand voice across the app and outbound touchpoints.

Moving from an app to a dialer

Standalone apps have an adoption cost, someone has to decide to download yet another app before they get any value from it. The direction being pursued was to make hiRobin a default Android dialer instead. As the default dialer, Robin isn't a separate destination, it's just how the phone already works: make your own calls normally, or hand a call or a task to Robin from the exact same interface. That removes the adoption step entirely and reframes what hiRobin is, from an app you have to remember to open, to the way calling already happens.

Three hiRobin dialer screens on Android phones: the side menu with Pause Robin, the Quick Calls and call log with Add Instruction and Make New Call buttons, and a call log with each call expanded to show its actions.

Scale

Roughly 1 million downloads within 6 to 9 months, growing to 160,000 to 200,000-plus active users. Signup-to-activation conversion moved from roughly 30 to 40% to over 80% following the OTP fix and onboarding rebuild, and D7 retention peaked at 74%.

Where it stands

Role and product work spanned January to October 2026. The results, the conversion fix, the growth to 200K-plus active users, the hiring and monetization scope, and the dialer strategy, stand on their own regardless of what came after.

At a glance

What it is

An AI phone assistant that handles inbound and outbound calls via call forwarding, not a chatbot, not a generic voice assistant

Role

Founding Product Designer

Tenure

Jan 2026 to Oct 2026

Downloads

~1 million within 6 to 9 months

Active users

160,000 to 200,000+

Conversion

~30 to 40% to ~80%+ signup-to-activation

D7 retention

23% to a peak of 74%

Strategic direction

Moving from a standalone app toward becoming a default Android dialer