Two-person mobile studio · Bengaluru
A two-person mobile studio. Agents do the volume work. Then a person reviews the architecture, tests on real phones, and signs off on security.
The floor
Open any desk to see the job and the human who checks it.
Human · cabin
Mobile across all four stacks: native iOS, native Android, KMP, Flutter.
Builder at Keenai Global, shipping features end to end, backend through testing, with AI, in production wealth-tech. He owns architecture review here, and shares code review with Tanya.
Owns
Architecture review, code review, design review, the release cut.
Human · cabin
Native Android and KMP, and the iOS side too.
Software Engineer 2 at Motive, where she owns releases. Here she owns the product spec, real-device QA, and the security, privacy and ASO review. She shares code review with Sahib.
Owns
Spec, code and design review, device QA, security and privacy, ASO, release cut.
Agent · desk 1
Turns the brief into a written spec: scope, constraints, and what the thing deliberately won't do.
Checked by
Product spec review
Agent · desk 2
Layouts, states, and the empty and error screens everyone else forgets.
Checked by
Design review
Agent · desk 3
Writes the code. Fast, and far more of it than a person would.
Checked by
Code review
Agent · desk 4
Writes the tests around the edges you'd have shipped without. The agent writes them and runs them. A person judges whether they test what matters.
Checked by
QA on real devices
Agent · desk 5
Scans dependencies, permissions, and what the app collects and where it goes. The agent runs the scan. A person reads the result and decides what to do about it.
Checked by
Security and privacy review
Agent · desk 6
Reads every diff for what a tired human misses at 1am.
Checked by
Code review
Agent · desk 7
Watches crashes and performance in production and files the ticket before your users do.
Checked by
Ship approval
The empty chair
This chair stays empty. No agent decides that something is ready for your users. A person does, every release.
The full pipeline
Spec Writer, Designer, Programmer, Test Engineer, Security Auditor, Reviewer and Release Watcher.
Where they've worked
Keenai Global
2025 – now
Motive
2023 – 2025
smallcase
2020 – 2023
Cleartrip
2019 – 2020
TheGeekDogs
2020 – now
Motive
2024 – now
HSBC
2021 – 2023
Naskay Technologies
2020 – 2021
Design review and the release itself stay with Sahib and Tanya: both read the layouts before build, and both names go on every release.
That's the difference between generated software and shipped software.
01
Sahib decides the shape before a line is written: what the data looks like, where the boundaries are, what this has to survive in two years. Agents are good at filling in a structure and bad at choosing one.
02
Every change is read by a person before it merges, and both of us are on this one. Not skimmed for style. Read for whether it does what it claims.
03
Tanya runs the build on physical phones, not just an emulator. Slow networks, low battery, older Android versions, the states people actually hit.
04
Tanya reads what the app collects, where it goes, what's stored and what's exposed, against what we told your users we'd do.
Proof
Pocket Manager has been in the Play Store since 2020, maintained by the same pipeline and the same review gates. 4.3 stars from 24 reviews, 1,000+ downloads, and it still gets updates five years on.
Milan, a wedding planner, followed it into the Play Store, built through the same pipeline and stopped at the same gates. Neither one was given a date before it went out. A person decided when each was ready.
Open Pocket Manager in the Play Store Opens a new tab.Read how it gets built
Two shipped, both in the store
Pocket Manager
Milan
The track shows both products and where each one actually is: Pocket Manager shipped in 2020 and still updated, Milan in the Play Store beside it. Neither was given a date before it went out, and nothing next will be. A date we miss in public is worse than no date at all.
The first thing we hand over is a working build you can open on your phone. Not a deck, not a Figma walkthrough, not a status update. Everything after that is iteration on a thing that already runs.
One project at a time. Fixed scope or a retainer, both work here. It starts with a short written brief rather than a call, and a first MVP is a few days from the go-ahead.
One email, a real reply from one of us, and a straight answer about whether we're the right fit. If we're not, we'll say so.