SiteOps AI

AI construction site copilot that turns multilingual daily reports into structured work logs, tasks, material alerts, delays, finance context, and manager messages.

Comment

SiteOps AI: AI Copilot for Real Construction Sites

The Problem

Construction crews run on quick voice notes, scribbled messages, and fast field updates. But important stuff gets lost all the time: how many workers actually showed up, what got finished today, which materials are missing, why things got delayed, who’s responsible, and what needs to happen tomorrow.

It gets even messier when people speak different languages — Russian, Hebrew, English. A foreman might report in Russian, the manager thinks in Hebrew, and the owner wants everything in English. That’s exactly why we built SiteOps AI.

What It Does

SiteOps AI is a real native Android and iOS app made for construction teams.

You just type or dictate a daily report in Russian, Hebrew, or English. GPT-5.6 turns that raw, messy update into a clean, structured summary that includes:

  • Site and date
  • Number of workers
  • Work completed
  • Floors and sections
  • Materials used
  • Materials still missing
  • Delays and reasons
  • Who’s responsible
  • Possible financial impact
  • Tomorrow’s tasks
  • Ready-to-send messages for managers

The app also has dashboards, site overviews, finance tracking, task lists, checklists, notifications, photo reports, offline mode with sync, and proper user roles (owner, manager, foreman, accountant, worker).

How We Built It

We went with React Native + TypeScript (no Expo) so we could have real native capabilities on both Android and iOS.

The backend is an Express API running on Railway with PostgreSQL. It handles login, sessions, report syncing, push notification tokens, AI analysis, transcription, and multilingual support.

We use OpenAI’s GPT-5.6 to turn natural language field reports into structured data while keeping all the important details (quantities, floor numbers, names, invoice numbers, dimensions) intact. Reports stay understandable no matter which language each team member prefers.

How Codex Helped

Codex was basically our co-founder during Build Week. It helped us with architecture decisions, building native screens, multilingual UI, role-based auth, fixing Android quirks, connecting the Railway backend, setting up PostgreSQL, implementing report sync, improving checklists and tasks, writing the README, and preparing the whole Devpost submission.

It was also great for spotting runtime bugs, reviewing screenshots, and quickly iterating on the MVP while we were testing on real devices.

Challenges We Faced

The hardest part was making the app feel like an actual construction tool instead of just another AI wrapper. We had to connect reports, sites, tasks, finances, checklists, and notifications into one smooth workflow.

Multilingual support was another big one. It’s not enough to translate the interface — the actual reports need to be readable by everyone in their own language.

Voice input brought its own headaches too — dealing with native permissions, different device capabilities, and platform differences.

What We Learned

We realized the real value of AI on construction sites isn’t just chatting. It’s turning chaotic field communication into reliable operational data: clear tasks, delay alerts, missing materials, responsibilities, and financial signals.

We also learned that a solid MVP needs both strong AI and proper business foundations — user accounts, roles, backend persistence, sync, and realistic testing scenarios.

What’s Next

We’re planning to add:

  • Production push notifications (FCM/APNs)
  • Cloud photo storage
  • Better offline conflict handling
  • Signed builds for Android and iOS
  • Monitoring and analytics
  • Deeper integrations with finance and project management tools

Our goal is simple: make SiteOps AI the go-to AI copilot that actually helps construction teams capture, understand, and act on what’s happening on site every day.

Built With

Share this project:

Updates