Skip to content

Awaisali36/custom-ai-project-management-system

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Custom AI Project Management System

Built for an Interior Architecture Studio — Dubai

Type   Built In   Saving



What This Is

A fully delivered, production-running project management system built for True Build Design — an interior architecture and project management studio in Dubai working on complex residential renovations for homeowners and investors.

True Build Design operates at the intersection of design intent and delivery precision. Their work covers everything from concept through to site supervision — drawings, material selections, approvals, shop drawings, and the detailed coordination that keeps a project from getting diluted by time, budget, and miscommunication.

That level of operational complexity demands systems that match it. This is that system.

"Clarity early. Decisions documented. Craft protected. Timelines respected." — True Build Design



The Problem

Interior architecture studios managing multiple live renovation projects face a specific operational challenge: every project generates a continuous stream of submissions, RFIs, design change notices, variation orders, vendor coordination, client approvals, and meeting action items — all of which must be tracked, escalated, and documented with precision.

The market solution — enterprise PM software — doesn't fit:

  • Procore, Autodesk Build, PlanGrid: $300–$800/user/month → $36K–$96K annually
  • Built for large-scale construction, not interior architecture workflows
  • Rigid processes that override the studio's own methodology
  • Vendor lock-in with no customisation capability

The manual alternative — spreadsheets and email — breaks at scale:

  • 20+ hours per week lost to admin across submission tracking, RFI follow-up, and meeting documentation
  • No SLA enforcement — approvals sit unreviewed for weeks
  • No audit trail for design decisions and client acknowledgments
  • Estimated $48,000/year in lost productive time


The System

A 6-database relational architecture built in Notion, managing the complete project lifecycle from client onboarding through to handover — across 15+ simultaneous projects.

┌─────────────┐
│   CLIENTS   │  ← Homeowners, investors, project sponsors
└──────┬──────┘
       │
┌──────▼──────┐
│   PROJECTS  │  ← 15+ live projects, health-monitored
└──────┬──────┘
       ├─────────────────┬──────────────────┐
       ▼                 ▼                  ▼
┌─────────────┐   ┌─────────────┐   ┌─────────────┐
│   RECORDS   │   │   ACTIONS   │   │   VENDORS   │
│             │   │             │   │             │
│ Submissions │   │ Tasks from  │   │ Suppliers,  │
│ RFIs · DCNs │   │ meetings,   │   │ contractors │
│ SIs · VOs   │   │ approvals   │   │ performance │
│ Minutes     │   │ escalations │   │ scoring     │
└─────────────┘   └─────────────┘   └─────────────┘
       │
┌──────▼──────┐
│  CONTACTS   │  ← Team members, clients, vendor contacts
└─────────────┘

What Runs Without Manual Intervention

Process What Happens Automatically
Submission Intake Reference number generated, Drive folder created, client acknowledgment sent
RFI Tracking 5-day SLA monitored continuously, escalated at breach — nothing waits in an inbox
Meeting Minutes AI extracts action items, creates tracked tasks, assigns owners and deadlines
Approval Routing PM → Design Lead → sign-off chain with PDF output and client notification
Daily Digest Priority-ranked open items delivered before the team's day begins
Project Health Continuous monitoring — flags overdue critical items before they become delays
Vendor Scoring Performance tracked automatically across response times and quality issues
Client Notifications Automated updates with acknowledgment tracking via webhook
Approval Reminders Nudges sent at Day 3, Day 7, Day 10 — nothing sits unreviewed
Executive Dashboard Weekly consolidated view of all project health and team output
Document Generation Registers, reports, and compliance documents generated on demand
Cross-Project Intelligence Recurring RFI themes identified across all active projects


Results

Metric Outcome
Software cost replaced $36K–$96K/year → $1,200/year
Weekly admin hours eliminated 20+ hours
Annual productivity recovered $48,000
Missed submissions or lost approvals Zero
Project delays from approval lag Reduced 50%+
Simultaneous projects managed 15+
Record types with custom lifecycles 8+
Specialised views across the system 15+
System ownership Full — client owns everything


How It Was Built

This system was not built through traditional software development. It demonstrates a new model for delivering custom enterprise tooling using AI.

Phase 1 — Architecture Design Strategic sessions to design the complete system — 6 databases, relationship logic, workflow triggers, approval hierarchies, and edge case handling. Every decision documented before anything was built.

Phase 2 — System Specification A 40-page master specification written covering every database property, relationship rule, automation requirement, and business logic condition. The complete operational blueprint.

Phase 3 — Autonomous Build An AI coding agent executed the entire build autonomously via Notion API — creating all 6 databases, their properties, views, relationships, and connections in 12 hours.

Phase 4 — Automation Layer 12 automated workflows deployed connecting the system to email, file storage, PDF generation, AI meeting processing, and webhook endpoints.

Traditional Dev This Approach
Build time 60–100 hours 12 hours
Build cost $3,000–$15,000 Under $150
Time saved 90%+
Cost saved 97%+
Ownership Vendor/agency Full client ownership


Technology

Layer Technology
Core Platform Notion (6 databases, 15+ views)
Automation 12 workflow pipelines
System Architecture ChatGPT
System Specification Claude
Build Execution Claude Code (autonomous agent)
Meeting Intelligence Groq LLM
Document Generation api2pdf
File Management Google Drive
Client Communications Gmail API
Acknowledgment Tracking Webhooks


Built by Trilles AI

This system was designed and delivered by Awais Ali, CEO & Co-Founder of Trilles AI.

If your studio or firm is managing complex projects across multiple clients and needs a system built precisely around how you work — not how a SaaS vendor thinks you should — this is what we build.

Connect on LinkedIn   Visit Trilles AI   Email



Built with precision · Powered by Trilles AI · www.trillesai.com

About

Enterprise-grade custom PM system built for architecture & construction. Replaces Procore at 1/30th the cost. 6-database architecture, 12 automated workflows, built in 12 hours using AI-assisted development. Saves $35K–$95K annually.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors