Audette Setup

Audette Setup

Let's get you up and running with a live demonstration that will:

• ✅ Test all functionality end-to-end
• 🏗️ Create a real example project
• 📊 Generate demonstration data
• 📄 Produce a sample report

Time: 5-10 minutes Result: Complete working example you can learn from

How This Works

I'll walk you through each capability by actually doing it with demo data:

1. System checks - Verify prerequisites
2. Workspace setup - Create demo project folder
3. Building creation - Add a real building profile
4. Energy data - Generate/extract 12 months of utility data
5. Equipment survey - Document building systems
6. RAG indexing - Index demo documents (if available)
7. Report generation - Create decarbonization roadmap

At each step, I'll:

• 🎯 Explain what we're doing
• ⚙️ Execute the actual function
• ✅ Verify it worked
• 💡 Show you the results

Step-by-Step Execution

Step 1: Pre-Flight Checks ✈️

What I'm doing: Verifying system prerequisites

Checks:

1. Audette MCP connected (for building profiles)
2. Public APIs accessible (OSM, Nominatim)

Execute checks:

# System tools
pdftotext -v 2>&1 | head -1
python3 --version

# Public APIs
curl -s "https://nominatim.openstreetmap.org/search?format=json&q=Seattle+WA" \
  -H "User-Agent: audette-skills/1.0 (https://github.com/soapboxbuild/audette-skills)" | jq -r '.[0].lat'

MCP test: Call list_customer_accounts() to verify Audette connection.

If checks fail:

• MCP error: Check claude_desktop_config.json, restart Claude Desktop
• API timeout: Check internet connection

On success: ✅ Proceed to workspace creation

Step 2: Create Demo Workspace 📁

What I'm doing: Setting up a complete project structure for "Green Tower Office Building"

Execute:

Create workspace at /tmp/audette-demo-{timestamp} with structure:

audette-demo-{timestamp}/
├── .audette-config.json
├── utility-bills/
│   └── (will create demo bill)
├── equipment/
│   └── (will create demo survey)
├── documents/
│   └── (will create demo PCA)
└── reports/
    └── (will generate roadmap)

Trigger: workspace-setup skill

User choice:

"I'll use fabricated demo data. Which building type would you like to explore?"

• A) Large Office Building (50,000 sq ft, urban, 1985)
• B) Multifamily Residential (100 units, suburban, 1970)
• C) Retail Center (75,000 sq ft, strip mall, 1995)
• D) Specify your own (provide address and details)

Default if no answer: Option A (Large Office)

Result:

• ✅ Workspace created
• ✅ .audette-config.json initialized
• ✅ Linked to your Audette account
• 💡 Show config file contents

Step 3: Create Building Profile 🏢

What I'm doing: Creating a real building in your Audette account with calculated GFA

Demo building (Option A - Large Office):

• Name: Green Tower Office Building
• Address: 1200 First Avenue, Seattle, WA 98101
• Type: office
• Year Built: 1985
• Floors: 12

Execute:

1. Geocode address → Get GPS coordinates via Nominatim
2. Query OSM footprint → Find building outline via Overpass API
3. Calculate GFA → Use Shoelace formula for polygon area × floors
4. Create in Audette → Call create_building() MCP tool

Trigger: audette-create-building skill OR direct MCP call

Result:

• ✅ Building created with building_uid and building_model_uid
• ✅ GFA calculated from OSM (or estimated ~50,000 sq ft)
• ✅ Added to workspace config
• 💡 Show building summary

Step 4: Generate Energy Data 📊

What I'm doing: Creating 12 months of realistic utility data

Demo data approach:

Option 1: Fabricate realistic data Generate synthetic utility bills based on building type and climate:

• Monthly electricity: 35,000 - 55,000 kWh (office baseline with seasonal variation)
• Monthly natural gas: 800 - 2,500 GJ (heating season peaks)
• Costs: $0.12/kWh electricity, $0.30/GJ gas (Seattle rates)
• Dates: Last 12 complete months

Option 2: User provides real bills

"Would you like to use fabricated demo data, or do you have real utility bills to extract?"

• Fabricated → I'll generate realistic synthetic data
• Real PDFs → Point me to your utility-bills folder

Execute (fabricated path):

1. Generate CSV data:

month,electricity_kwh,electricity_cost,gas_gj,gas_cost
2024-01,52000,6240,2400,720
2024-02,48000,5760,2200,660
2024-03,42000,5040,1800,540
... (12 months total)

2. Create demo PDF bill (one sample for show):

SEATTLE CITY LIGHT
Account: DEMO-12345
Service Period: January 2024

Electric Charges
Usage: 52,000 kWh
Total: $6,240.00

3. Trigger: audette-energy-data skill
4. Upload to Audette: Call add_building_utility_data MCP tool

Result:

• ✅ 12 months of energy data validated
• ✅ Saved to workspace
• ✅ Sample PDF created in utility-bills/
• 💡 Show monthly consumption chart (simple table)

Step 5: Equipment Survey 🔧

What I'm doing: Documenting building HVAC and major systems

Demo equipment (Large Office baseline):

Heating:

• 2× Natural gas boilers (500 MBH each, installed 2010)
• Hot water distribution with variable speed pumps

Cooling:

• 3× Rooftop air conditioning units (40 tons each, installed 2015)
• Direct expansion cooling

Ventilation:

• 4× Rooftop air handling units with economizers
• Demand control ventilation (CO2 sensors)

Domestic Hot Water:

• 1× Natural gas water heater (75 gallon, commercial)

Lighting:

• 60% LED (retrofitted 2020)
• 40% fluorescent T8 (original)

Controls:

• Basic BMS (Building Management System)
• Programmable thermostats by zone

Execute:

Option 1: Fabricate survey Generate complete equipment survey JSON and markdown report

Option 2: Interactive questionnaire

"Would you like to:"

• A) Use fabricated demo equipment (typical 1985 office, partial retrofits)
• B) Answer a few questions to customize the equipment (5-10 questions)
• C) Extract from a document (if you have a PCA/CNA PDF)

Trigger: audette-equipment-survey skill OR submit_equipment_survey() MCP

Result:

• ✅ Complete equipment inventory
• ✅ Mapped to Audette schema
• ✅ Saved as equipment/green-tower-survey.json
• 💡 Show system summary table

Step 6: RAG Document Indexing 🔍 (Optional)

What I'm doing: Creating searchable document index for semantic queries

Check dependencies:

python3 -c "import sentence_transformers; print('Available')" 2>&1

If available:

1. Create demo documents:

   • documents/pca-summary.pdf - Property condition assessment excerpt
   • documents/engineer-report.pdf - Mechanical systems report

2. Generate sample content:

PROPERTY CONDITION ASSESSMENT - GREEN TOWER
Prepared by: Building Analytics Inc.
Date: March 2024

HVAC SYSTEMS
The building heating is provided by two natural gas-fired
boilers located in the basement mechanical room. Each boiler
has a rated capacity of 500 MBH and was installed in 2010.
The boilers are in good condition with regular maintenance.

Expected remaining useful life: 8-10 years.
Replacement cost: $45,000 per unit.

The cooling system consists of three rooftop packaged air
conditioning units...

3. Trigger: rag_ingest MCP tool
4. Test query: "What type of heating system is installed?"

If not available: ⚠️ Skip - RAG is optional. Note: "Install Soapbox RAG (via rag_ingest MCP) for document search"

Result (if available):

• ✅ Documents indexed with embeddings
• ✅ Test query returns relevant excerpt
• 💡 Show sample Q&A

Step 7: Generate Decarbonization Roadmap 📄

What I'm doing: Creating a complete HTML report with interactive charts

Execute:

1. Prepare report data:

   • Baseline emissions from energy data
   • Equipment inventory for measure recommendations
   • Placeholder carbon reduction measures:
     • Boiler replacement with heat pumps (-180 tCO2e/yr, $120k)
     • LED lighting completion (-25 tCO2e/yr, $15k)
     • Rooftop solar PV installation (-100 tCO2e/yr, $200k)
     • HVAC controls upgrade (-40 tCO2e/yr, $30k)

2. Trigger: report skill with template decarbonization-roadmap

3. Generate artifacts:

   • HTML report with:
     • Executive summary
     • Baseline emissions breakdown (Scope 1, 2, 3)
     • Measure economics table
     • Implementation timeline (Gantt chart)
     • Cash flow analysis (line chart)
     • Emissions trajectory (area chart)
   • All charts use Chart.js
   • "Print to PDF" button included

Result:

• ✅ Full decarbonization roadmap created
• ✅ Interactive HTML artifact displayed
• ✅ Saved to reports/green-tower-roadmap.html
• 💡 Open artifact and explore charts

Step 8: Review & Next Steps 🎯

What we accomplished:

File tree:

audette-demo-{timestamp}/
├── .audette-config.json (linked to Audette account)
├── utility-bills/
│   ├── energy-data.csv (12 months)
│   └── sample-bill-2024-01.pdf
├── equipment/
│   └── green-tower-survey.json
├── documents/
│   ├── pca-summary.pdf
│   └── .rag-index/ (if RAG enabled)
└── reports/
    └── green-tower-roadmap.html ⭐

Try these next:

1. Explore the report artifact

   • Click on charts to see interactivity
   • Use Print → Save as PDF
   • See investment-grade styling

2. Modify the demo data

   • Edit energy-data.csv with different values
   • Re-run audette-energy-data skill
   • Regenerate report to see changes

3. Ask questions (if RAG enabled)

   • "What's the condition of the boilers?"
   • "When were the RTUs installed?"
   • "What's the replacement cost for HVAC?"

4. Start a real project

   • Use this demo as a template
   • Replace with actual building data
   • Follow the same workflow

Demonstration Complete! 🎉

You now have a complete working example of the entire Audette Orchestrator workflow.

What you learned:

• ✅ How to structure projects
• ✅ How to onboard buildings with GFA calculation
• ✅ How to handle energy data
• ✅ How to document equipment
• ✅ How to generate professional reports

Ready for real work?

Typical workflow for actual projects:

1. Week 1: Data Collection

   - Create workspace
   - Add building (use actual address)
   - Gather utility bills (ask client)
   - Collect property assessments
   

2. Week 2: Analysis

   - Extract energy data from PDFs
   - Interview building operator (equipment survey)
   - Index documents with RAG
   - Review in Audette platform
   

3. Week 3: Modeling

   - Run baseline calibration in Audette
   - Identify reduction measures
   - Run financial analysis
   - Review measure economics
   

4. Week 4: Delivery

   - Generate final report
   - Add client logo
   - Export to PDF
   - Present to stakeholders
   

Try next:

• system-details — Generate building systems report
• audette-energy-data — Process real utility bills

Need help? Ask me:

• "How do I onboard my building?"
• "Extract these utility bills"
• "Generate a BPS compliance report"
• "What's the difference between report templates?"

Cleanup

I've created demo data. Would you like to keep it or clean up?

Keep it if:

• You want a reference example
• You're learning the workflows
• You'll use it for testing

Delete it if:

• You're ready to start real projects
• You want a clean slate
• Disk space is limited

To clean up:

rm -rf /tmp/audette-demo-*

Note: The building "Green Tower Office Building" was created in your Audette account. You can delete it from the Audette dashboard if desired, or keep it as a sandbox.

Welcome Aboard! 🚀

You're now equipped to create world-class decarbonization roadmaps with AI.

The Audette Orchestrator helps you:

• ⚡ Work faster - Automation saves hours
• 🎯 Work accurately - Fewer manual errors
• 📊 Work professionally - Beautiful deliverables

Let's decarbonize buildings together! 🌱

Have questions? Just ask - I'm your AI assistant for building decarbonization.