Asheerty

Asheerty is for families whose history is still on paper or in one person's memory.

A family starts with one name and adds relatives, photos, dates and their stories. Others can join through a shared link to look at the tree or help build it. I made the app in Arabic, with right-to-left layouts from the beginning, rather than treating the language as something to add later.

The tree can stay in the app, but it can also become a framed poster. Each framed poster has a QR code that opens the same living tree, keeping the artwork connected to the family's record. This is why I built the mobile app, poster editor and order process together.

A tree is useful only if people can find their way around it. I built a canvas with pan, zoom and collapsible branches, using Shopify's React Native Skia to draw the connections and render only what's in view. There is also a card view and search for someone who wants to find a person without exploring the whole diagram.

Adding a relative opens an editor for their name, relationship, life dates, photos and story. Quick additions help a family grow the tree gradually, and an offline queue keeps edits made without a connection. Relatives can propose changes, which the owner reviews before accepting them. The tree can be shared without giving everyone the same editing access.

The AI Tree Builder reads a photograph, but the family still checks the result.

Asking a family to type out a tree they already have on paper creates a lot of repeated work. I built an import flow where they photograph the chart and tap its root. The backend uses OpenAI vision to read names and notes, map the relationships and reconstruct a draft tree.

One pass is not enough to trust the result. A separate review looks for missed people and branches, and server checks catch broken links and cycles before the draft reaches the app. Uncertain names and relationships stay marked for review. AI helps with the transcription, while the family remains responsible for deciding what is correct.

The mobile app collects the order. The dashboard is where I prepare the artwork.

The print flow lets a customer choose the size, frame, copies and recipients, then follow the order from preview to delivery. Behind that flow, I built an Arabic dashboard and a poster editor that works in centimetres. The tree, frame, background, QR code and notes all belong to the same layout.

The editor can draw the family as branches and leaves, concentric rings or cards connected from top to bottom. I added generated variations, alignment tools, clash warnings, undo and redo, and saved snapshots so a layout can be worked on without losing earlier decisions. It exports SVG, HTML, PDF and PNG, and sends a watermarked preview to the customer's app.

Nothing goes to print before the customer accepts the preview and the first payment is confirmed. Each order keeps its approved artwork, including for a reprint. Version checks prevent a stale save from overwriting a newer layout, and the dashboard offers only the action that belongs to the order's current stage.

I built the API with TypeScript on Bun, Express, Prisma and PostgreSQL. It handles family records, photos, shared trees and print orders. Redis provides caching, MinIO stores photos and assets, and NotifyX handles push notifications. These pieces support the same journey, from adding a relative to receiving a poster.

Shared links also need to work outside the app. I made readable share pages with a title and image for WhatsApp and Facebook, and an invitation page that opens the tree or sends a relative to the right app store. The Arabic website includes a sample tree people can try in their browser. I also connected campaign links and Android install attribution to account creation so I can follow where sign-ups come from.

I maintain both native projects, use Fastlane for store uploads and host the app's JavaScript update service myself. Backend tests cover the API, while health checks and metrics help me operate it after release.

I built a WhatsApp assistant with n8n on top of WaSphere, an open source WhatsApp gateway. My work includes hosting the gateway, updating its WhatsApp engine, fixing video message handling and building the assistant around it.

The assistant answers questions about Asheerty in the customer's language, including Iraqi Arabic. It does not invent account, payment or order information. When a question needs private information or a person, it hands the conversation over and stays quiet. Message tracking prevents duplicate replies, and delivery records let me see whether a response was actually sent. This gives routine questions an answer while keeping a person involved where the assistant cannot know enough.

Making the app also meant making its marketing videos. I built Studio in a week in September 2026, using two Windows PCs of my own to run render jobs through ComfyUI. From the browser I can write a brief, add image and audio references, send a job and follow its progress.

A Director flow breaks a storyboard into clips with timing, narration and references. I review those clips before rendering them, and finished clips can be assembled into one MP4 in the browser with ffmpeg.wasm. A separate pipeline makes short tutorial videos from HTML pages and checks their text placement before encoding.

Studio and the support assistant are part of the same work as the app and poster editor. Building Asheerty on my own has meant building the tools I need to explain it, support it and keep it running.

A bug on a screen can be fixed with an update. A mistake that has been printed, framed and shipped costs real money. That made review part of the product, from checking an imported family tree to approving the final artwork. The app keeps the family record alive, while the order process makes sure the version that leaves the printer is the one the customer accepted.