ReplyRail: Local Order Follow-ups

Actual local prototype, synthetic records. Editing and history persisted after reload; no messaging or AI inference.

Problem

Small community sellers can lose track of orders and promised replies across conversations. ReplyRail explores a local queue of unresolved promises. This is a problem hypothesis, not validated merchant adoption.

Working prototype

Orders are ranked by promised deadline. Users can add orders, search and filter statuses, update the next action, inspect history, and copy a follow-up draft. Nothing sends automatically. BHD values use integer fils. Backup imports validate every record before replacing the workspace.

Video demonstrations

Five-minute product demonstration for Next Founders.

Short working prototype demonstration for WarriorHacks. The short demo edits and retimes the actual captured footage, with local synthetic narration. Both videos disclose AI assistance and use only synthetic records.

Five-minute downloadable video

Download or view the five-minute demonstration. The video shows the actual local app, editing and persistence, code architecture, business hypothesis and limitations. It uses sampled browser-viewport capture with original wall-clock timing and local synthetic narration. Frames hold during capture gaps. All records are synthetic; no private customer information or credentials are shown. The downloadable MP4 is also available through the five-minute YouTube link above.

Build and provenance

Created October 5-6, 2026 for OFFGRID and already submitted there. Reuse for Next Founders and WarriorHacks is explicitly disclosed. This is not a renamed Clearledger entry. Built with substantial Codex AI assistance, JavaScript modules, Node.js, localStorage and Lucide 0.468.0. No hosted model or paid API.

Verification

26 domain tests passed without failures or skips. Syntax checks passed. Browser checks verified synthetic loading, editing, history and persistence after reload. The video fully decodes, is five minutes long and contains non-silent audio. These checks do not establish production security or merchant effectiveness.

Business hypothesis

The target audience is small community sellers managing orders manually. A free single-device version could be followed by a future paid encrypted-backup or team edition. BHD 2-5 per month is a price experiment, not proven willingness to pay. First conduct consented interviews and a usability pilot, measuring weekly reuse and resolved overdue promises. Competing alternatives include spreadsheets, reminders and existing order tools. No paying customers, revenue or rewards are claimed.

Run it

Download the source archive from the public repository. With Node 22+, run npm ci, npm test, npm run check, then npm start. Open http://127.0.0.1:3240 locally and use synthetic records. This is not a publicly hosted app.

Limits and scaling

Single device, 1,000 orders and a 2 MB import cap. Storage and backups are not encrypted. Use customer labels, not private records. No WhatsApp integration, automatic messages, remote sync, AI inference or payments. Larger local storage, encrypted backups, multi-user authorization, accessibility and recovery testing are future work, not implemented features.