Saved $24,000 a year and built seven automations to scale from 117 to 919 homes

THE AUTOMATION ARCHITECT AWARD - PROJECT DESCRIPTION

Visionary Creator of AI-Driven Automations

The Vision

I see automation differently than most people. It's not about saving a few clicks—it's about building intelligent systems that think, extract, analyze, and act without human intervention.

But here's the key: I don't build alone. AI is my thought partner.

Every automation I create starts with a conversation with Claude. I describe the problem. We brainstorm solutions together. Claude helps me see edge cases I'd miss, suggests architectural approaches, and thinks through the logic with me. Then we build it together—Claude Code and VS Code as my development environment, with AI helping me write, debug, and refine.

As Head of Organizational Development at Fifth Avenue Properties—a homebuilder scaling from 117 to 919 home starts this year—I've become the automation architect for our entire operation. Our BHAG is 5,000 homes per year. Using n8n, Claude Code, VS Code, and a growing stack of APIs, I'm building the infrastructure that will let us scale 8x this year and eventually 40x—without scaling our administrative burden at the same rate.

This is what I've built. This is what's coming.

Automation #1: The $24,000/Year BuildPro Integration

The Problem: Our construction management software vendor wanted $2,000/month for an API connector to sync BuildPro data with Airtable.

The Automation I Built:

BuildPro → Scheduled Email Reports → n8n Webhook

→ Parse Excel Attachment → Transform Data

→ Airtable API → Update Job Records

→ Trigger Downstream Dashboards

How It Works:

1. Configured BuildPro to automatically email reports on a schedule

2. n8n workflow monitors incoming email, extracts Excel attachment

3. Parses the spreadsheet data and transforms it for Airtable schema

4. Pushes data via Airtable API to update construction timelines

5. Dashboards automatically refresh with current build times

The Impact: $24,000/year saved—with more flexibility than the vendor solution. I control the logic. I can modify it anytime. No vendor lock-in.

Tools: n8n, Airtable API, Email parsing

Automation #2: Meeting Intelligence Pipeline

The Problem: Four standing meetings every week. Notes scattered. Action items lost. No institutional memory.

The Automation I Built:

Workflow #1: Read AI → Notion Importer (5 nodes, Active)

Read AI Webhook → Extract Meeting Data

→ Create Meeting in Notion → Append Full Content

→ Log Success

Workflow #2: Notion → AI Summarizer → Airtable (15+ nodes, Active)

Schedule Trigger (Daily) → Query Notion Database

→ Loop Over Meetings → Get Page Blocks

→ Convert to Transcript → Check Length

→ Claude API: Generate Summary → Parse Sections

→ Get Existing Properties → Build Conditional Update

→ Check If Update Needed → Update Notion Page

→ Prepare Airtable Data → Route by Meeting Type

→ Create Exec/Production Meeting Record → Mark Synced

→ Log Results

How It Works:

1. Meeting occurs → Read AI automatically captures transcript

2. Webhook fires → n8n creates Notion page with full content

3. Daily scheduled workflow loops through unprocessed meetings

4. Claude API generates structured summary (decisions, action items, key points)

5. Routes to appropriate Airtable table based on meeting type

6. Marks as synced to prevent reprocessing

The Impact: Hours saved weekly. Zero manual processing. Complete institutional memory captured automatically. Team gets action items with full context.

Tools: n8n, Read AI webhooks, Notion API, Claude API, Airtable API

Automation #3: Receipt Extractor

The Problem: Expense receipts buried in email. Manual data entry into expense tracking. Time wasted, receipts missed.

The Automation I Built:

Email Inbox Monitor → Identify Receipt Emails

→ Extract Attachment/Content → Parse Receipt Data

→ Airtable API → Create Expense Record

→ Categorize & Tag

How It Works:

1. n8n monitors email inbox for receipt patterns

2. Extracts receipt data (vendor, amount, date, category)

3. Creates structured record in Airtable expense tracking system

4. Auto-categorizes based on vendor recognition

The Impact: No more lost receipts. No more manual entry. Expense data flows automatically into our financial tracking.

Tools: n8n, Email parsing, Airtable API

Automation #4: Customer Progress Notifications

The Problem: Homebuyers want to know what's happening with their build. Manual updates are inconsistent and time-consuming.

The Automation I Built:

Construction Task Completed (Airtable checkbox)

→ Trigger Workflow → Pull Customer Data

→ Generate Personalized Email

→ Send via Outlook → Log Communication

How It Works:

1. Construction team clicks checkbox when milestone is completed

2. Automation triggers immediately

3. Pulls customer contact info and project details

4. Generates personalized email: "Here's what just happened on your home, and here's what's coming next"

5. Sends automatically, logs the communication

The Impact: Customers stay informed without manual effort. Consistent communication at every milestone. Happy customers, less administrative burden.

Tools: n8n, Airtable API, Outlook/Email API

Automation #5: Weekly Sales Reports

The Problem: Manual report generation every week. Pulling data, formatting, distributing.

The Automation I Built:

Schedule Trigger (Weekly) → Query Airtable Sales Data

→ Aggregate & Calculate Metrics

→ Generate Report → Distribute to Stakeholders

The Impact: Reports go out automatically every week. No manual work. Leadership gets consistent data on schedule.

Tools: n8n, Airtable API

Automation #6: Utility Provider Extraction Agent

The Problem: As a homebuilder, we deal with hundreds of utility accounts across multiple providers. Each has a different portal, different login, different interface. Extracting balances manually takes hours every month.

The Automation I Built:

Scheduled Trigger → AI Agent Queue

→ For Each Utility Provider:

→ Navigate to Portal → Authenticate

→ Extract Account Balance → Extract Due Date

→ Capture to Spreadsheet

→ Generate Payment Summary

→ Flag Exceptions for Human Review

How It Works:

1. AI agent logs into each utility provider website

2. Navigates the portal (each one is different)

3. Extracts the amount owing and due date

4. Populates a spreadsheet for payment processing

5. Flags any exceptions or errors for human review

The Impact: Hours of manual portal-hopping eliminated. One spreadsheet with all utility balances ready for payment processing.

Tools: n8n, AI Agent, Browser automation

Automation #7: Flight Booking Assistant

The Problem: Travel booking is time-consuming. Searching options, comparing prices, getting approval, completing the booking—30+ minutes per trip.

The Automation I Built:

Teams Message: "I need to fly to Vancouver next Tuesday"

→ AI Agent Parses Request

→ Searches Flight Options → Compares Prices/Times

→ Presents Top 3 Options in Teams

→ User Selects & Authorizes

→ Agent Completes Booking → Confirms in Teams

→ Creates Calendar Event → Logs Expense

How It Works:

1. Initiated via Teams message with travel request

2. AI agent parses the request (destination, dates, preferences)

3. Searches flight options and compares prices/times

4. Presents top options back in Teams for selection

5. User authorizes payment

6. Agent completes booking, confirms, and logs everything

The Impact: 30-minute task becomes a 2-minute decision. Human stays in the loop for authorization, AI handles the research and execution.

Tools: n8n, AI Agent, Microsoft Teams API

Automation #8: AI Email Draft Assistant (In Refinement)

The Problem: Email takes hours. Reading, thinking, drafting responses. It's necessary but repetitive.

The Automation I Built:

New Email Received → n8n Webhook Triggered

→ Parse Email Content & Context

→ Claude API: Analyze Intent & Generate Draft Response

→ Save Draft for Review

→ Human Reviews & Sends (or Edits)

How It Works:

1. Incoming emails trigger the workflow

2. AI analyzes the email content, sender context, and intent

3. Generates a draft response matching my communication style

4. Draft is ready for my review—I approve, edit, or rewrite

5. Human stays in the loop for final send

Current Status: Under active development. The core automation works—drafts are being generated. Refining the AI's understanding of context and tone to get responses closer to perfect on the first pass.

Why It Matters: Even at 80% accuracy, this saves significant time. I'm not starting from a blank page—I'm editing a draft. As the AI learns my style, the edits get smaller.

Tools: n8n, Email API, Claude API

IN DEVELOPMENT: The Next Level

These are the AI agent automations I'm currently building—taking automation from simple triggers to intelligent systems that navigate, decide, and act.

Coming Soon #1: AI Building Permit Agent

The Vision: Automatically fill out building permit applications for new home construction.

The Architecture:

New Project Created in Airtable

→ AI Agent Pulls Project Data (lot, specs, community)

→ Navigates to Municipal Permit Portal

→ Fills Form Fields Intelligently

→ Flags for Human Review → Submit on Approval

Why It Matters: Permit applications are repetitive but require accuracy. An AI agent can pre-fill 90% of the work, letting humans focus on review and exceptions. At 919 homes this year—and 5,000 at our BHAG—this isn't a nice-to-have. It's essential infrastructure.

Coming Soon #2: Automation Problem Collection System

The Vision: Crowdsource automation opportunities from the entire organization.

The Architecture:

Airtable Form Sent to All Staff

→ "What repetitive task is eating your time?"

→ Submissions Collected → AI Analyzes & Prioritizes

→ Automation Architect Reviews

→ Build Queue Created → Track Development

Why It Matters: The best automation ideas come from the people doing the work. This system turns every employee into an automation scout—surfacing problems I'd never see from my seat. Instead of guessing what to automate next, I'll have a prioritized list of real pain points from across the organization.

Coming Soon #3: SOP Execution Workflows

The Vision: SOPs that don't just document—they execute. Using SweetProcess integrated with n8n to track completion and trigger downstream actions.

The Architecture:

SOP Step Completed → Trigger Next Action

→ Notify Relevant People → Update Status

→ If Final Step → Archive & Report

Why It Matters: SOPs that track themselves. Accountability built in. No more wondering if the process was followed.

The Technical Foundation

APIs I've Connected:

- Airtable API (central data hub)

- Notion API (knowledge base)

- Read AI webhooks (meeting capture)

- Claude API (AI summarization and analysis)

- Microsoft Graph API (SharePoint, Outlook, Teams)

- Email parsing (receipts, BuildPro reports)

Development Environment:

- n8n (workflow orchestration)

- Claude Code (AI-assisted development)

- VS Code (code editing and debugging)

How I Work — AI as My Thought Partner:

This is the part most people miss. The tools are just tools. The real power is using AI as a thought partner throughout the entire automation lifecycle:

Ideation: When I face a repetitive problem, I don't just jump to building. I talk it through with Claude. "Here's what's painful. Here's what I'm thinking. What am I missing?" The AI helps me see possibilities I wouldn't find alone—like the BuildPro email workaround that saved $24K/year.

Planning: Before I touch n8n, I map out the architecture with Claude. We walk through the data flow, identify edge cases, think about error handling. "What happens if the email doesn't have an attachment? What if the API rate limits us?" This planning prevents rebuilding later.

Implementation: Claude Code and VS Code are where ideas become reality. I describe what I need, AI helps write the code, I test and refine. When something breaks, I debug with Claude—pasting error messages, walking through logic together.

Optimization: After an automation runs for a while, I review it with Claude. "Here's what's working. Here's what's failing. How do we make it better?" The AI Email Draft Assistant is in this phase right now—working, but being refined through iteration.

This isn't just clicking nodes together. This is architectural thinking—with an AI thought partner who never gets tired of talking through problems.

Current State & Trajectory

Active Workflows: 7 production workflows + 1 in refinement Weekly Executions: ~10 (and growing) Annual Savings (Documented): $24,000+ (BuildPro alone) Monthly Savings (Estimated): $2,000+ (data retrieval efficiency)

What's Coming:

- AI agent for building permit applications (919 homes/year)

- Automation problem collection from all staff

- SOP execution tracking

- Expanded customer communication automation

Complete Tool Stack

Automation Platform: n8n AI Development: Claude Code, VS Code APIs Connected: Airtable, Notion, Read AI, Claude, SharePoint, Outlook, Teams, Obsidian Data Hub: Airtable Knowledge Base: Notion (company), Obsidian (personal)

Key Learnings

1. AI as thought partner transforms what's possible — I couldn't build these automations alone. Brainstorming with Claude, planning with Claude, debugging with Claude—this collaboration lets me punch way above my weight class.

2. Start with the painful repetition — Find the tasks people do over and over. That's where automation has the biggest impact.

3. AI agents are the next level — Simple trigger-action automations are table stakes. AI agents that can navigate, decide, and act are the future.

4. Build vs. Buy pays off — The $24K/year BuildPro savings proved that creative automation beats vendor solutions.

5. Design for scale — I'm building for 919 homes this year, but architecting for 5,000. Every automation needs to handle 40x volume without breaking.

6. Human-in-the-loop for high stakes — AI agents research and recommend; humans authorize and approve. Best of both worlds.

7. Iterate in public — The email drafter isn't perfect yet. That's okay. Ship, learn, improve. Waiting for perfection means never shipping.

Relevant URLs

fifthaveproperties.com, fifthavehomes.com

Closing

I don't just use automation tools. I architect systems—with AI as my thought partner every step of the way.

From the first spark of an idea to the final debugging session, Claude is there. We brainstorm together. We plan together. We build together. We fix things together.

Every workflow I build is designed to eliminate repetitive work, capture institutional knowledge, and free humans to do what humans do best—think, decide, and create.

We're scaling from 117 to 919 homes this year. Our BHAG is 5,000. The only way to get there is with intelligent systems that scale faster than headcount. With AI agents on the horizon, I'm not just automating tasks—I'm building the infrastructure for a company that can build 5,000 homes a year.

This is what an Automation Architect does. This is how we scale.