Last Planner System · Construction Controls

Every Commitment.
Every Constraint.
Tracked Daily.

A remote project controls platform built on the Last Planner System — field reports collected every evening via Telegram, WhatsApp, or Email; an AI agent analyzes them overnight; your team validates at morning review before the first tool hits steel.

Open LPS → About LPS Methodology
5
LPS planning levels in one app
3
Report channels — Telegram, WhatsApp, Email
4
n8n automation templates, one active at a time
◈
AI agent analyses every evening, human validates every morning

Daily Operations

How the System Runs Every Workday

The cycle is simple and repeatable. Field teams report in plain language — the system handles the structure, the AI handles the analysis, and your planners handle the decisions.

STEP 01 · EVENING
Field Reports In
Foremen send daily progress via Telegram, WhatsApp, or Email. Photos, GPS, weather attached automatically by the n8n workflow.
STEP 02 · NIGHT
AI Agent Analyzes
The OpenClaw-style HQ agent reads all reports, cross-references constraints and PPC, and prepares summaries and recommended actions.
STEP 03 · MORNING
Human Validates
Project Controls opens the morning review queue. Each AI analysis can be validated, overridden, or flagged — nothing moves forward without human sign-off.
STEP 04 · DAY
Plan Updated
Validated reports feed the Weekly Work Plan and Lookahead Board. Constraints are logged, PPC is updated, and the cycle begins again at 5 PM.

App Modules

Eight Tabs, One Workflow

Every tab maps to a level of the Last Planner System — from the long-range phase plan down to yesterday's field report.

⬡
Dashboard
Live KPIs: floor progress visualizer, PPC trend, crew counts by trade, active constraints. The one screen to open every morning.
🔀
Phase Pull Planning
Swim-lane board with promises and requests per trade. Add tasks, assign responsible parties, and map hand-offs across Survey, Design, Engineering, MEP, and Construction.
📅
3-Week Lookahead
Constraint-based pull scheduling across three weeks. Trade filter, status color codes, and constraint flags surface problems before they hit the field.
✓
Weekly Work Plan
Day-by-day commitment grid per trade with MON–FRI status indicators. Live PPC calculation against committed tasks.
📈
PPC Tracking
Dedicated Percent Plan Complete dashboard: trend chart, variance breakdown by cause, monthly performance, benchmark comparison, and CSV export.
📥
Report Inbox
All incoming field reports in one view — channel badge, AI analysis status, and the Human-in-the-Loop VALIDATE button for the morning review workflow.
◈
AI Agent
Chat directly with the HQ construction agent. Demo mode simulates responses. Live mode calls Claude Sonnet with your API key. Swap for a local Ollama agent at any time.
⚙
Automation Hub
Four n8n workflow templates — activate one at a time, view the node diagram, export the JSON, or copy the import URL to your live n8n instance.

n8n Automation

Four Templates, One Channel at a Time

The Automation Hub lets you pick which workflow powers the field reporting channel. Each template is a real exportable n8n JSON — swap them without touching code.

🤖 AI Daily Report Telegram trigger → message classifier → OpenAI analysis → Google Sheets log + Email notification. The default template for this demo.
📸 Photo Progress Tracker Scheduled Google Drive scan → batch image processor → Gemini AI construction analysis → Sheets append + stakeholder email.
📋 Task + Photo Reports Telegram bot with command switch → user registration → task CRUD → photo + GPS upload → Google Sheets sync. Full project management via chat.
🔄 Multi-Channel Manager WhatsApp + Telegram webhook → message router → task management engine → status checker → notification engine. Handles both channels simultaneously.

AI Architecture

Demo Agent vs Real Local Agent

In Demo mode the agent simulates responses using pre-built answers tuned to the Pacific Tower project context. Switch to Live mode and the Settings drawer lets you wire in the real Claude API — or replace it with a local Ollama / OpenClaw agent running on your own infrastructure.

Demo Mode (default)

No API key needed. Realistic pre-built responses cover daily report summaries, constraint analysis, PPC risk, and report validation guidance. Ideal for stakeholder demos without live credentials.

Live Mode (Settings ⚙)

Enter your Anthropic API key in Settings. The agent calls Claude Sonnet with the full project context on every message. Or replace the fetch URL with http://localhost:11434/api/chat to run Ollama locally — same OpenAI-compatible format.

Ready to start

Open the Controls App

Pacific Tower · 32-Floor Mixed Use · New York City · Week 10 of 44

Open App → Read Methodology