Tushar Nath

Digital transformation · APAC

All products

AI product

SnapTask

Turns screenshots into tracked actions: whatever arrives, on any app, a screenshot of it is read by AI and lands in a calendar-shaped tracker with the task and its due date already filled in.

Status
In development
Built with
Expo React Native · TypeScript · Expo Router · Supabase · OpenAI via a secure backend

The case study

The problem

Obligations no longer arrive as tasks. They arrive as a WhatsApp message glanced at in a queue, a bill photographed and forgotten, a date mentioned halfway down a notice, each one carrying a deadline and none of them wearing a label. Memory takes custody of them all, and memory is the one system guaranteed to drop something. SnapTask stands on a single observation: whatever the platform, whatever the format, a screenshot can hold it. So the screenshot becomes the front door. Capture the moment something is promised, and the promise should walk itself into a calendar: named, dated, and impossible to lose the way a chat message is lost.

The approach

Instead of an integration for every platform, SnapTask asks for the one gesture every phone owner already performs without thinking. A screenshot goes up; AI on a secure backend reads what it says; a calendar-shaped tracker fills itself in, task and due date already written. The plumbing is deliberately familiar: Expo React Native and TypeScript, Supabase underneath, intelligence kept behind the backend and never in the app, the same foundations already carrying its older sibling Niyavo.

The result

The concept is settled: screenshot in, tracked action out. The build is carrying it forward, and by August 2026 it had reached its middle stages.

What it taught

There is a quieter experiment running inside this one. Niyavo settled on Claude Code, so SnapTask deliberately began on Codex, an assistant new to its builder, because a second product is exactly where a second tool should be learned: real stakes, familiar territory, honest comparison. The clearest reading of an AI tool comes from pushing a real product through it, and that is the reading being taken here: capabilities, habits and limits, logged one build at a time, so the next project starts with a toolbox already in hand.

Source: Confirmed directly by the owner · confirmed 2026-08-20

More products