Atapata

Digital business cards for Pakistan

Founder and sole engineer · April 2025 – present · HydraLabs, incubated at NIC NED

The problem

Business cards in Pakistan still get printed, handed over, and lost. The information on them goes stale the moment someone changes jobs or numbers. And the person who receives the card has to either type it into their phone or keep a stack in a drawer.

The digital alternatives that exist mostly assume a US or European user: pricing in dollars, payment methods that don't work locally, and no answer for the fact that most of the cards you'll receive in Karachi are still paper.

Atapata is built for the actual situation: you have a digital card, and the people you meet mostly don't.

What it does

You create a card in the app and share it three ways: tap an NFC card against someone's phone, show a QR code, or send a WhatsApp link. The person receiving it doesn't need the app.

The Atapata card dashboard showing a founding member card with share, QR, NFC, edit and new actions, view and save counts, and a recent activity feed. — The card dashboard: share, QR, NFC and edit, with view and save counts per card.

The Atapata share screen showing a large QR code with options to share via WhatsApp, copy link, email, NFC tag or save the QR. — Sharing: QR, WhatsApp, link, email or NFC tag. An offline mode shares the QR without a connection.

An incoming Atapata card saved to the inbox, with a prompt asking whether to share your own card back. — Receiving a card. It saves straight to the inbox, then offers to send yours back.

Because the card is a live profile rather than a file, changing your job title updates every card you've ever shared. Nobody has a stale copy.

For the paper cards you still collect, the app scans them. Point the camera at a card, and OCR pulls the name, title, company, phone, and email into a contacts inbox. You can attach a voice note to any contact while the conversation is fresh, and it tags where you met them.

The Atapata camera aimed at a printed business card, with flash, filters and shutter controls. — Capturing a paper card in the native document scanner.

A progress screen reading your card at 85 percent, with steps for uploading the image, reading text, extracting fields and matching to your fields. — OCR in progress: upload, read text, extract fields, match to your fields.

The review contact screen with fields for event, date, location, meeting notes and an in-progress voice memo recording. — The review step before saving: where you met, meeting notes, location, and a voice memo recorded on the spot.

The inbox, mixing scanned paper cards with Atapata profiles.

You can also order a physical NFC card, delivered in Pakistan, that points at your profile.

A physical Atapata NFC business card held in a hand. — The physical NFC card, delivered in Pakistan.

The app in use.

How it's built

The mobile app is React Native on Expo, running the New Architecture. The API is NestJS with Prisma over PostgreSQL, deployed on Railway. The web app is Next.js on Vercel. Files live in Cloudflare R2, background jobs run through BullMQ on Redis, and transactional email goes through Resend.

Card scanning uses Claude Haiku for OCR. Payments go through Safepay. Analytics through PostHog, error tracking through Sentry.

Releases ship over the air through EAS Update with fingerprint-based runtime versioning, so most changes reach users without an app store review.

The decisions that mattered

OCR is the paywall, because OCR is the cost. Every AI scan costs money per call, and it is the only part of the product whose cost scales with how much people use it. Everything else is fixed to serve, so gating any of it would be an arbitrary tax on the parts that are already paid for. Scanning is what the paid tier gates. The pricing follows the unit economics instead of fighting them.

Rolling windows, not calendar weeks. Scan quotas reset on a trailing 30-day window rather than resetting on the 1st. Calendar-anchored limits create a spike of usage at month end and a dead zone at the start, and they feel arbitrary to the user who signed up on the 28th.

Entitlements are computed when you read them. There's no cron job flipping subscription states at midnight. An entitlement is derived from grant records at query time, which means no drift between what the database thinks and what's true, and no window where a job failure leaves someone locked out of something they paid for. Concurrent writes take a row lock so two simultaneous scans can't both slip past a quota.

The scanner had to be native. The first version cropped camera frames manually with expo-camera. It worked on iOS and was quietly broken on Android, where the measurement call returns zeros and the crop lands nowhere near the card. The fix was a native document scanner plugin, which is TurboModule-only and forced the app onto the New Architecture. That was a larger migration than the bug seemed to warrant, and it was still the right call.

Founding member templates are hidden, not locked. They were originally visible to everyone with a lock icon. On device that read as a nag rather than something exclusive, so non-members don't see them at all now.

Business model

Pricing is set in rupees and paid through a local processor. A dollar-denominated subscription is a non-starter for most of this market.

Where it is

Running as an open TestFlight beta with Android in parallel.

The Founding Members programme covers the first 200 people to install, sign up, and create a card. All 200 get Pro free for six months; the first 50 also get a physical NFC card. As of the last count, 26 members had activated, creating 30 cards, sharing them 167 times, and scanning 59 paper cards.

A How Atapata works screen listing guided walkthroughs for creating a card, sharing it, scanning a paper card, writing to an NFC tag, organising the inbox, voice notes and ordering a physical card. — In-app walkthroughs covering each part of the product, for beta users arriving without any onboarding call.

Public launch is set for 11 November 2026.

← Back home