Jeep Hillhouse (RMI Construction) - RMI Job Tracker

The Problem

RMI ran multiple residential and commercial job sites with plumbers, electricians, carpenters, and tile setters, but job information was scattered across texts, calls, and notebooks. Workers arrived without door codes, client details, or clear tasks. Assignments were often missed in message threads, and completed work had limited accountability records.

The owner also needed a low-friction system designed for phones and real field conditions. Typing in complex software was not practical for workers with dirty hands on active job sites.

The Turning Point

Weekly planning consumed Sunday nights, and wrong-site trips plus missed details were recurring. Coordination through dozens of text conversations was no longer scalable. A field-first, phone-first, voice-enabled workflow became mandatory.

The Solution

1. Single job dashboard with address, client info, door code, assignees, and status.
2. Voice commands for creating jobs, assigning workers, and setting urgency.
3. Per-worker views so each crew member sees only relevant tasks.
4. @mentions with instant push and email notifications (no app install required).
5. Field photo uploads with timestamp and uploader metadata.
6. Complete activity logs for who changed what and when.
7. Magic-link login for frictionless access from notification emails.
8. TV dashboard mode for office-wide live visibility.
9. Natural language search for jobs and assignments.
10. High-contrast, large-target interface tuned for older users and field use.

The Results

• Communication: Daily coordination volume dropped from 50+ texts to 5-10.
• Wrong-site trips: Address and door-code confusion was effectively eliminated.
• Planning time: Sunday planning dropped from ~2 hours to ~30 minutes.
• Task reliability: Assignment tracking and mentions reduced missed tasks.
• Accountability: Timestamped photos and activity logs created verifiable records.
• Scale: RMI managed about 30% more jobs without adding office staff.

Tech Stack

• Frontend: React Native with Expo (web, iOS, Android from one codebase)
• Backend: Node.js, Express, PostgreSQL
• AI: Claude API for natural-language command parsing
• Voice: Browser Web Speech API for transcription
• Notifications: Web Push (PWA), email, Twilio SMS
• Hosting: Replit with auto-scaling
• Build time: 6 weeks from concept to production

The Human Detail

"I am not a tech guy, and I do not want to learn software. But with this app, I just talk to it. Create a job, assign Bob, send him a message. My crew gets notified, they know where to go, and I can see everything from my phone in my truck. That is the difference."

The Takeaway

This project showed that modern AI-powered field software can be adopted daily by teams that do not want complex training. Voice-first interaction is practical today, and a single app can replace text chaos, spreadsheets, photo-email threads, and sticky-note scheduling.