AI Design Assistant - Complete Training Course

<h4>Welcome to the AI Design Assistant Training Course</h4> <p>This comprehensive guide will teach you how to leverage artificial intelligence to streamline your MEP engineering workflow.</p>

<h4>What is the AI Design Assistant?</h4> <p>The AI Design Assistant is a built-in feature of J∆S Engineering Suite that provides intelligent assistance for HVAC and MEP design tasks. It uses large language models (LLMs) to understand your questions in natural language and provide contextually relevant answers based on industry standards, building codes, and engineering best practices.</p>

<h4>Key Capabilities:</h4> <ul> <li><strong>Preliminary Load Estimation:</strong> Get quick cooling/heating load estimates based on building type and square footage</li> <li><strong>Equipment Selection Guidance:</strong> Receive recommendations for chillers, air handlers, boilers, and other equipment from real manufacturers</li> <li><strong>Code Compliance Assistance:</strong> Ask questions about ASHRAE 90.1, 62.1, Title 24, IPC, IMC, and other codes</li> <li><strong>Design Checklists:</strong> Generate building-type-specific design checklists automatically</li> <li><strong>Energy Optimization:</strong> Get strategies for improving energy efficiency</li> <li><strong>Troubleshooting:</strong> Diagnose comfort complaints and system issues</li> </ul>

<h4>Why Use AI in Engineering Design?</h4> <ul> <li><strong>Speed:</strong> Get instant answers instead of searching through manuals</li> <li><strong>Consistency:</strong> Ensure design approaches align with industry standards</li> <li><strong>Learning:</strong> Understand the reasoning behind recommendations</li> <li><strong>Productivity:</strong> Focus on engineering decisions, not information gathering</li> </ul>

<h4>Important Disclaimer:</h4> <p>AI assistance is a tool to augment your engineering expertise, not replace it. Always verify AI recommendations against current code requirements and use professional judgment for final design decisions. The AI provides guidance based on general best practices - your specific project may have unique requirements.</p>

<h4>Choose Your AI Provider</h4> <p>J∆S Engineering Suite supports multiple AI providers, giving you flexibility in how you access AI assistance. Choose based on your privacy requirements, budget, and internet connectivity.</p>

<h4>Free AI Options:</h4> <table class="table table-bordered"> <thead><tr><th>Provider</th><th>Cost</th><th>Privacy</th><th>Internet Required</th><th>Best For</th></tr></thead> <tbody> <tr><td><strong>Ollama (Local)</strong></td><td>100% Free</td><td>Complete - runs on your PC</td><td>No (offline capable)</td><td>Privacy-sensitive projects, offline use</td></tr> <tr><td><strong>Google Gemini</strong></td><td>Free tier (1,500 req/day)</td><td>Data sent to Google</td><td>Yes</td><td>High quality, generous free tier</td></tr> <tr><td><strong>Groq</strong></td><td>Free tier available</td><td>Data sent to Groq</td><td>Yes</td><td>Very fast responses</td></tr> </tbody> </table>

<h4>Paid AI Options:</h4> <table class="table table-bordered"> <thead><tr><th>Provider</th><th>Cost</th><th>Quality</th><th>Best For</th></tr></thead> <tbody> <tr><td><strong>Claude (Anthropic)</strong></td><td>~$3-15 per million tokens</td><td>Excellent reasoning</td><td>Complex engineering questions</td></tr> <tr><td><strong>OpenAI GPT</strong></td><td>~$2-10 per million tokens</td><td>Very good general purpose</td><td>Wide range of tasks</td></tr> </tbody> </table>

<h4>Simulated Mode (Offline Fallback):</h4> <p>When no AI provider is configured or available, the assistant falls back to "Simulated" mode. This mode uses built-in responses for common engineering queries - no internet or API key required. While less flexible than true AI, it still provides:</p> <ul> <li>Preliminary load estimates based on rules of thumb</li> <li>Standard equipment recommendations</li> <li>Basic code compliance summaries</li> <li>Design checklists by building type</li> </ul>

<h4>Recommendation for New Users:</h4> <p>Start with <strong>Ollama</strong> if you want completely free, private AI that works offline. Start with <strong>Google Gemini</strong> if you want the easiest setup with high quality responses. Use <strong>Simulated mode</strong> if you just want quick estimates without any setup.</p>

<h4>Why Ollama?</h4> <p>Ollama is the recommended free option because it runs entirely on your computer - your project data never leaves your machine. This section walks through complete setup.</p>

<h4>System Requirements:</h4> <ul> <li>Windows 10/11, macOS 10.15+, or Linux</li> <li>8GB RAM minimum (16GB recommended for larger models)</li> <li>10-50GB disk space depending on models</li> <li>No GPU required (but recommended for faster responses)</li> </ul>

<h4>Step 1: Install Ollama</h4> <ol> <li>Visit <a href="https://ollama.ai" target="_blank">https://ollama.ai</a></li> <li>Click "Download" and select your operating system</li> <li>Run the installer (Windows: .exe, Mac: .dmg)</li> <li>Follow the installation prompts</li> </ol>

<h4>Step 2: Download an AI Model</h4> <p>Open a terminal/command prompt and run:</p> <pre class="bg-dark text-light p-3 rounded"><code># Recommended for most users (7B parameters, ~4GB) ollama pull llama3.2

# Smaller, faster option (3B parameters, ~2GB) ollama pull llama3.2:3b

# Larger, more capable (13B parameters, ~8GB) ollama pull llama3.2:13b

# Alternative: Mistral (good for technical content) ollama pull mistral</code></pre>

<h4>Step 3: Start the Ollama Server</h4> <pre class="bg-dark text-light p-3 rounded"><code>ollama serve</code></pre> <p>Keep this terminal window open. Ollama runs on http://localhost:11434 by default.</p>

<h4>Step 4: Configure J∆S Engineering Suite</h4> <ol> <li>Open J∆S Engineering Suite</li> <li>Navigate to Tools, then AI Design Assistant</li> <li>In the Provider dropdown, select "Ollama (Local, Free)"</li> <li>The assistant will automatically connect to localhost:11434</li> </ol>

<h4>Optional: Environment Variables</h4> <pre class="bg-dark text-light p-3 rounded"><code># Windows (in System Environment Variables) OLLAMA_URL=http://localhost:11434 OLLAMA_MODEL=llama3.2</code></pre>

<h4>Troubleshooting Ollama:</h4> <ul> <li><strong>"Connection refused":</strong> Make sure <code>ollama serve</code> is running in a terminal</li> <li><strong>"Model not found":</strong> Run <code>ollama pull llama3.2</code> to download the model</li> <li><strong>Slow responses:</strong> Try a smaller model like llama3.2:3b, or close other applications</li> <li><strong>Out of memory:</strong> Use a smaller model or add more RAM</li> </ul>

<h4>Cloud AI Provider Setup</h4> <p>If you prefer cloud-based AI or need more capable models, you can configure Google Gemini, Groq, Claude, or OpenAI.</p>

<h4>Google Gemini Setup (Free Tier):</h4> <ol> <li>Visit <a href="https://makersuite.google.com/app/apikey" target="_blank">Google AI Studio</a></li> <li>Sign in with your Google account</li> <li>Click "Create API Key"</li> <li>Copy the API key</li> <li>Set the environment variable or enter directly in J∆S Engineering Suite</li> <li>Install the required package: <code>pip install google-generativeai</code></li> </ol> <p><strong>Free Tier Limits:</strong> 1,500 requests/day, 32,000 tokens/minute</p>

<h4>Groq Setup (Free Tier):</h4> <ol> <li>Visit <a href="https://console.groq.com" target="_blank">Groq Console</a></li> <li>Create an account</li> <li>Navigate to API Keys and create a new key</li> <li>Set the GROQ_API_KEY environment variable</li> <li>Install the required package: <code>pip install groq</code></li> </ol> <p><strong>Why Groq?</strong> Groq uses custom hardware for extremely fast inference - responses in under 1 second.</p>

<h4>Anthropic Claude Setup (Paid):</h4> <ol> <li>Visit <a href="https://console.anthropic.com" target="_blank">Anthropic Console</a></li> <li>Create an account and add billing information</li> <li>Generate an API key</li> <li>Set the ANTHROPIC_API_KEY environment variable</li> <li>Install the required package: <code>pip install anthropic</code></li> </ol>

<h4>OpenAI Setup (Paid):</h4> <ol> <li>Visit <a href="https://platform.openai.com" target="_blank">OpenAI Platform</a></li> <li>Create an account and add billing information</li> <li>Generate an API key</li> <li>Set the OPENAI_API_KEY environment variable</li> <li>Install the required package: <code>pip install openai</code></li> </ol>

<h4>Entering API Keys in the Application:</h4> <p>You can also enter API keys directly in the AI Design Assistant window:</p> <ol> <li>Select your provider from the dropdown</li> <li>Enter your API key in the "API Key" field</li> <li>The key is used for the current session only (not saved to disk)</li> </ol> <p><strong>Security Note:</strong> API keys entered in the application are only stored in memory during your session. For permanent configuration, use environment variables.</p>

<h4>Six Specialized Assistant Modes</h4> <p>The AI Design Assistant has six specialized modes, each optimized for different engineering tasks. Switching modes adjusts the AI system prompt and response style.</p>

<h4>1. Project Manager Mode</h4> <p><strong>Best for:</strong> Overall project guidance, system selection decisions, design approach</p> <p><strong>Example queries:</strong></p> <ul> <li>"What HVAC system type should I use for a 50,000 SF office building?"</li> <li>"What are the key design considerations for a laboratory project?"</li> <li>"Help me plan the MEP design phases for this healthcare facility"</li> </ul>

<h4>2. Equipment Selector Mode</h4> <p><strong>Best for:</strong> Specific equipment recommendations with manufacturer/model suggestions</p> <p><strong>Example queries:</strong></p> <ul> <li>"What chiller do you recommend for 200 tons cooling?"</li> <li>"Compare Trane vs Carrier rooftop units"</li> <li>"What boiler efficiency should I specify for ASHRAE 90.1 compliance?"</li> </ul> <p>Provides real manufacturer equipment from Trane, Carrier, Daikin, York, Greenheck, etc.</p>

<h4>3. Code Compliance Mode</h4> <p><strong>Best for:</strong> Questions about building codes and standards</p> <p><strong>Example queries:</strong></p> <ul> <li>"What is the minimum ventilation rate for an office per ASHRAE 62.1?"</li> <li>"Does my system need an economizer per ASHRAE 90.1?"</li> <li>"What are the Title 24 requirements for lighting controls?"</li> </ul>

<h4>4. Load Calculator Mode</h4> <p><strong>Best for:</strong> Estimating heating, cooling, and ventilation loads</p> <p><strong>Example queries:</strong></p> <ul> <li>"What is the cooling tonnage for my building?"</li> <li>"Estimate the ventilation CFM for 250 occupants"</li> <li>"What heating BTU/hr do I need for a warehouse in Chicago?"</li> </ul>

<h4>5. Energy Optimizer Mode</h4> <p><strong>Best for:</strong> Energy efficiency strategies and sustainability guidance</p> <p><strong>Example queries:</strong></p> <ul> <li>"How can I reduce energy consumption in this building?"</li> <li>"What is the ROI on VFDs for my pumps?"</li> <li>"Should I use heat recovery for this project?"</li> </ul>

<h4>6. Troubleshooter Mode</h4> <p><strong>Best for:</strong> Diagnosing comfort complaints and system issues</p> <p><strong>Example queries:</strong></p> <ul> <li>"Why is the second floor always too hot?"</li> <li>"The VAV boxes are hunting - what should I check?"</li> <li>"Chiller short-cycling - possible causes?"</li> </ul>

<h4>How to Change Modes:</h4> <ol> <li>In the AI Design Assistant window, locate the "Assistant Mode" dropdown</li> <li>Select the mode that matches your current task</li> <li>The mode change takes effect immediately for your next query</li> </ol>

<h4>Why Project Context Matters</h4> <p>The AI provides much better recommendations when it understands your project. Always set the project context before asking detailed questions.</p>

<p>Without context, the AI gives generic answers. With context, it tailors recommendations to:</p> <ul> <li>Your specific building type (healthcare vs office vs laboratory)</li> <li>Your climate zone (affecting equipment sizing and efficiency requirements)</li> <li>Your building size (appropriate system types and capacities)</li> <li>Your occupancy (ventilation and load calculations)</li> </ul>

<h4>Project Context Fields:</h4> <table class="table table-bordered"> <thead><tr><th>Field</th><th>Description</th><th>Example</th></tr></thead> <tbody> <tr><td><strong>Building Type</strong></td><td>Primary function of the building</td><td>Office, Healthcare, Laboratory, Data Center</td></tr> <tr><td><strong>Square Footage</strong></td><td>Gross building area</td><td>50,000 SF</td></tr> <tr><td><strong>Location</strong></td><td>City and state</td><td>Los Angeles, CA</td></tr> <tr><td><strong>Climate Zone</strong></td><td>ASHRAE/IECC climate zone</td><td>3B (warm-dry)</td></tr> <tr><td><strong>Occupancy</strong></td><td>Number of people</td><td>250</td></tr> </tbody> </table>

<h4>Climate Zone Reference:</h4> <pre class="bg-dark text-light p-3 rounded"><code>Zone 1: Very Hot (Miami) Zone 2: Hot (Houston, Phoenix) Zone 3: Warm (Los Angeles, Las Vegas) Zone 4: Mixed (Seattle, New York, DC) Zone 5: Cool (Chicago, Boston, Denver) Zone 6: Cold (Minneapolis, Milwaukee) Zone 7: Very Cold (Duluth, Fargo) Zone 8: Subarctic (Fairbanks)

A = Humid, B = Dry, C = Marine</code></pre>

<h4>Building Type Implications:</h4> <ul> <li><strong>Healthcare:</strong> AI will consider ASHRAE 170, pressure relationships, 100% OA requirements</li> <li><strong>Laboratory:</strong> AI will consider fume hoods, high exhaust rates, diversity factors</li> <li><strong>Data Center:</strong> AI will consider ASHRAE TC 9.9, redundancy, PUE targets</li> <li><strong>Education:</strong> AI will consider classroom ventilation, acoustics, budget constraints</li> </ul>

<h4>Example: Impact of Context on Responses</h4> <p><strong>Without context:</strong> "What cooling system should I use?"</p> <p><em>AI response:</em> "Consider chillers, RTUs, or VRF depending on your needs..."</p>

<p><strong>With context (50,000 SF Healthcare, Climate 4A):</strong> "What cooling system should I use?"</p> <p><em>AI response:</em> "For a 50,000 SF healthcare facility in Climate Zone 4A, I recommend a water-cooled chiller plant with 100% outdoor air-capable air handling units. Healthcare facilities per ASHRAE 170 require specific pressure relationships and often 100% OA for critical areas..."</p>

<h4>Equipment Selection with AI</h4> <p>The AI Design Assistant can recommend specific equipment from major manufacturers. Here is how to get the most useful recommendations.</p>

<h4>Setting Up for Equipment Queries:</h4> <ol> <li>Set your project context (building type, size, location)</li> <li>Switch to "Equipment Selector" mode</li> <li>Ask about specific equipment types</li> </ol>

<h4>Effective Equipment Queries:</h4> <p><strong>Good:</strong> "What chiller do you recommend for my 50,000 SF office building?"</p> <p><strong>Better:</strong> "Recommend a water-cooled centrifugal chiller for 150 tons with high part-load efficiency"</p> <p><strong>Best:</strong> "Compare Trane CenTraVac vs Carrier AquaEdge for a 150-ton office application in Climate Zone 3B with emphasis on IPLV efficiency"</p>

<h4>Equipment Categories Available:</h4> <ul> <li><strong>Chillers:</strong> Trane CenTraVac, Carrier AquaEdge, York YVAA, Daikin Magnitude</li> <li><strong>Rooftop Units:</strong> Carrier WeatherExpert, Trane IntelliPak, Daikin RoofPak</li> <li><strong>Air Handlers:</strong> Trane M-Series, Carrier 39M, Daikin Modular</li> <li><strong>Boilers:</strong> Cleaver-Brooks, Lochinvar, Aerco</li> <li><strong>Cooling Towers:</strong> BAC, Evapco, SPX Marley</li> <li><strong>Fans:</strong> Greenheck, Twin City, Cook</li> <li><strong>Pumps:</strong> Bell and Gossett, Armstrong, Grundfos</li> <li><strong>VRF Systems:</strong> Daikin VRV, Mitsubishi City Multi, LG Multi V</li> </ul>

<h4>Understanding AI Equipment Recommendations:</h4> <p>When the AI recommends equipment, it typically provides:</p> <ul> <li><strong>Manufacturer and Model:</strong> Specific product line</li> <li><strong>Capacity Range:</strong> Available sizes</li> <li><strong>Efficiency Rating:</strong> IPLV, SEER2, HSPF2, etc.</li> <li><strong>Key Features:</strong> What makes this suitable for your application</li> <li><strong>Alternatives:</strong> Other options to consider</li> </ul>

<h4>Verify Before Specifying:</h4> <p>Always verify AI recommendations by:</p> <ol> <li>Checking manufacturer websites for current models</li> <li>Confirming availability in your region</li> <li>Reviewing actual submittal data</li> <li>Consulting with manufacturer representatives</li> </ol>

<h4>Navigating Building Codes with AI</h4> <p>The Code Compliance mode helps you navigate building codes and standards. Here is how to get accurate code guidance.</p>

<h4>Supported Codes and Standards:</h4> <ul> <li><strong>ASHRAE 90.1:</strong> Energy Standard for Buildings</li> <li><strong>ASHRAE 62.1:</strong> Ventilation for Acceptable Indoor Air Quality</li> <li><strong>ASHRAE 170:</strong> Ventilation of Healthcare Facilities</li> <li><strong>Title 24:</strong> California Energy Code</li> <li><strong>IPC:</strong> International Plumbing Code</li> <li><strong>IMC:</strong> International Mechanical Code</li> <li><strong>IFC:</strong> International Fire Code</li> <li><strong>UFC:</strong> Unified Facilities Criteria (DOD)</li> <li><strong>NFPA:</strong> Various fire protection standards</li> </ul>

<h4>Common Code Questions by Category:</h4> <h5>Ventilation (ASHRAE 62.1):</h5> <ul> <li>What is the minimum OA rate for [space type]?</li> <li>How do I calculate system ventilation efficiency?</li> <li>When is demand control ventilation required?</li> </ul>

<h5>Energy (ASHRAE 90.1):</h5> <ul> <li>What is the minimum chiller efficiency for [capacity]?</li> <li>When are economizers required?</li> <li>Is energy recovery required for my system?</li> </ul>

<h5>Healthcare (ASHRAE 170):</h5> <ul> <li>What are the pressure relationships for [room type]?</li> <li>What is the minimum ACH for [room type]?</li> <li>Can I recirculate air in [room type]?</li> </ul>

<h4>Example Code Compliance Query:</h4> <pre class="bg-dark text-light p-3 rounded"><code>You: What are the ASHRAE 62.1-2022 ventilation requirements for a conference room with 20 occupants and 400 SF?

AI: Per ASHRAE 62.1-2022 Table 6-1 for Conference/Meeting:

Breathing Zone Outdoor Air: - People component: 5 cfm/person x 20 = 100 cfm - Area component: 0.06 cfm/sf x 400 sf = 24 cfm - Total Vbz = 124 cfm

Zone Outdoor Air: Voz = Vbz / Ez - Ceiling supply, heating: Ez = 0.8, Voz = 155 cfm - Ceiling supply, cooling: Ez = 1.0, Voz = 124 cfm

Design Guidance: - Size for heating condition (155 cfm) - Consider CO2-based DCV for variable occupancy</code></pre>

<h4>Important Code Caveats:</h4> <ul> <li>Codes are updated regularly - verify current edition requirements</li> <li>Local amendments may modify model code requirements</li> <li>AHJ interpretations vary by jurisdiction</li> <li>Use the AI as a starting point, then verify with official documents</li> </ul>

<h4>Preliminary Load Estimates with AI</h4> <p>The Load Calculator mode provides preliminary heating and cooling load estimates. This is useful for early-stage budgeting and system selection, though detailed calculations should follow.</p>

<h4>What the AI Can Estimate:</h4> <ul> <li>Preliminary cooling tonnage</li> <li>Heating BTU/hr requirements</li> <li>Supply air CFM</li> <li>Ventilation requirements</li> <li>Load component breakdowns</li> </ul>

<h4>Rules of Thumb the AI Uses:</h4> <table class="table table-bordered"> <thead><tr><th>Building Type</th><th>SF/Ton (Cooling)</th><th>BTU/SF (Heating)</th></tr></thead> <tbody> <tr><td>Office</td><td>350-450</td><td>25-35</td></tr> <tr><td>Retail</td><td>300-400</td><td>25-40</td></tr> <tr><td>Restaurant</td><td>150-250</td><td>30-50</td></tr> <tr><td>Healthcare</td><td>250-350</td><td>35-50</td></tr> <tr><td>Laboratory</td><td>150-300</td><td>40-60</td></tr> <tr><td>Data Center</td><td>50-150</td><td>N/A</td></tr> <tr><td>Warehouse</td><td>800-1200</td><td>15-25</td></tr> </tbody> </table>

<h4>Typical Load Components:</h4> <pre class="bg-dark text-light p-3 rounded"><code>Cooling Load Components (typical office): - Envelope (walls, roof, windows): 25-35% - Solar (windows, skylights): 15-25% - People: 15-20% - Lighting: 10-15% - Equipment: 10-20% - Ventilation: 15-25%

Heating Load Components: - Envelope transmission: 60-80% - Infiltration: 10-20% - Ventilation: 10-30%</code></pre>

<h4>When to Use AI vs Detailed Calculations:</h4> <table class="table table-bordered"> <thead><tr><th>Use AI Estimates For:</th><th>Use Detailed Calcs For:</th></tr></thead> <tbody> <tr><td>Budgeting and fee proposals</td><td>Final equipment sizing</td></tr> <tr><td>Early system selection</td><td>Permit submittals</td></tr> <tr><td>Sanity checking detailed calcs</td><td>Title 24/90.1 compliance</td></tr> <tr><td>Quick "what-if" comparisons</td><td>Warranty documentation</td></tr> </tbody> </table>

<h4>Understanding AI Limitations</h4> <p>While the AI Design Assistant is a powerful tool, understanding its limitations is crucial for professional engineering practice.</p>

<h4>What the AI Does Well:</h4> <ul> <li>Providing quick preliminary estimates and rules of thumb</li> <li>Explaining code requirements and their rationale</li> <li>Suggesting equipment options from major manufacturers</li> <li>Generating design checklists and workflows</li> <li>Answering "what-if" questions quickly</li> <li>Providing troubleshooting approaches</li> </ul>

<h4>What the AI Cannot Do:</h4> <ul> <li><strong>Stamp drawings</strong> - AI output does not constitute professional engineering services</li> <li><strong>Guarantee code compliance</strong> - Always verify with current code editions</li> <li><strong>Know your specific project</strong> - Only knows what you tell it</li> <li><strong>Access real-time data</strong> - Cannot check current product availability or pricing</li> <li><strong>Replace detailed calculations</strong> - Estimates are approximate</li> <li><strong>Know local amendments</strong> - May not reflect jurisdiction-specific requirements</li> </ul>

<h4>When to Verify Manually:</h4> <h5>Always Verify:</h5> <ul> <li>Code requirements cited by the AI (check current edition)</li> <li>Equipment specifications (check manufacturer data)</li> <li>Sizing calculations for permit submittals</li> <li>Any safety-critical decisions</li> <li>Healthcare and laboratory requirements (high consequences)</li> </ul>

<h4>AI Hallucination Awareness:</h4> <p>Large language models can occasionally "hallucinate" - generating plausible-sounding but incorrect information. This is more likely when:</p> <ul> <li>Asking about very specific equipment models or part numbers</li> <li>Requesting exact code section numbers</li> <li>Asking about recent changes or updates</li> <li>Combining multiple unusual requirements</li> </ul>

<h4>Best Practices for Reliable Results:</h4> <ol> <li><strong>Cross-reference critical information</strong> - Use the AI as a starting point, then verify</li> <li><strong>Ask for sources</strong> - Request code sections or standard references</li> <li><strong>Provide context</strong> - More project details = better answers</li> <li><strong>Use simpler queries</strong> - Break complex questions into parts</li> <li><strong>Sanity check numbers</strong> - Do the values make intuitive sense?</li> </ol>

<h4>Professional Responsibility:</h4> <p>Remember that you, as the licensed professional, are responsible for the accuracy of your designs. The AI is a productivity tool that can help you work faster and consider more options, but it does not replace your engineering judgment, education, and experience.</p>

<h4>Data Privacy by Provider</h4> <p>Understanding how your data is handled by different AI providers is important, especially for confidential projects.</p>

<table class="table table-bordered"> <thead><tr><th>Provider</th><th>Data Location</th><th>Training Use</th><th>Retention</th></tr></thead> <tbody> <tr><td><strong>Ollama (Local)</strong></td><td>Your computer only</td><td>Never</td><td>None (deleted on session end)</td></tr> <tr><td><strong>Google Gemini</strong></td><td>Google servers</td><td>May be used (check policy)</td><td>Varies by API settings</td></tr> <tr><td><strong>Groq</strong></td><td>Groq servers</td><td>Check current policy</td><td>Varies</td></tr> <tr><td><strong>Claude</strong></td><td>Anthropic servers</td><td>Not used for training (API)</td><td>30 days (deletable)</td></tr> <tr><td><strong>OpenAI</strong></td><td>OpenAI servers</td><td>Not used for training (API opt-out available)</td><td>30 days (deletable)</td></tr> <tr><td><strong>Simulated</strong></td><td>Your computer only</td><td>Never</td><td>None</td></tr> </tbody> </table>

<h4>For Confidential Projects:</h4> <p>If working on projects with confidentiality requirements:</p> <ol> <li><strong>Use Ollama (Local)</strong> - Data never leaves your computer</li> <li><strong>Use Simulated mode</strong> - No external connections</li> <li><strong>Avoid including</strong> in queries: client names, specific addresses, proprietary system details, security-sensitive information</li> </ol>

<h4>What Data is Sent to Cloud Providers:</h4> <ul> <li>Your question/prompt text</li> <li>Project context you have set (building type, SF, location)</li> <li>Conversation history (last 10 messages)</li> <li>Your IP address and API key</li> </ul>

<h4>What is NOT Sent:</h4> <ul> <li>Project files from J∆S Engineering Suite</li> <li>Detailed model data</li> <li>Calculation results (unless you paste them in)</li> <li>Other files on your computer</li> </ul>

<h4>Government and Classified Projects:</h4> <p>For projects under ITAR, classified, or government restrictions:</p> <ul> <li><strong>Only use Ollama or Simulated mode</strong></li> <li>Do not use any cloud-based AI providers</li> <li>Verify your organization AI policy</li> </ul>

<h4>Recommendation by Project Type:</h4> <table class="table table-bordered"> <thead><tr><th>Project Type</th><th>Recommended Provider</th></tr></thead> <tbody> <tr><td>Standard commercial</td><td>Any provider acceptable</td></tr> <tr><td>Healthcare (PHI involved)</td><td>Ollama or Simulated</td></tr> <tr><td>Government/DOD</td><td>Ollama or Simulated only</td></tr> <tr><td>Competitive bid (confidential pricing)</td><td>Ollama preferred</td></tr> <tr><td>High-security facilities</td><td>Ollama or Simulated only</td></tr> </tbody> </table>

<h4>Hands-On Exercises</h4> <p>Practice using the AI Design Assistant with these exercises. Each builds on skills from previous sections.</p>

<h4>Exercise 1: Basic Setup and First Query</h4> <ol> <li>Open J∆S Engineering Suite</li> <li>Navigate to Tools, then AI Design Assistant</li> <li>Verify "Simulated (Offline)" is selected</li> <li>Type: "What HVAC systems are common for office buildings?"</li> <li>Review the response and note the level of detail</li> </ol>

<h4>Exercise 2: Setting Project Context</h4> <ol> <li>Set these project parameters: <ul> <li>Building Type: Office</li> <li>Square Footage: 75,000</li> <li>Location: Chicago, IL</li> <li>Climate Zone: 5A</li> <li>Occupancy: 400</li> </ul> </li> <li>Click "Set Project Context"</li> <li>Now ask: "What cooling tonnage do I need?"</li> <li>Compare this response to Exercise 1 - note how context improves answers</li> </ol>

<h4>Exercise 3: Changing Modes</h4> <ol> <li>Keep the same project context</li> <li>Change mode to "Equipment Selector"</li> <li>Ask: "What chiller do you recommend?"</li> <li>Change mode to "Code Compliance"</li> <li>Ask: "What efficiency does ASHRAE 90.1 require for my chiller?"</li> <li>Notice how responses adapt to the mode</li> </ol>

<h4>Exercise 4: Code Compliance Deep Dive</h4> <ol> <li>Stay in Code Compliance mode</li> <li>Try these queries: <ul> <li>"What is the outdoor air requirement for the break room?"</li> <li>"Do I need an economizer for a 15-ton RTU in Chicago?"</li> <li>"What is the maximum fan power allowance per ASHRAE 90.1?"</li> </ul> </li> <li>For each answer, note which code/standard is referenced</li> </ol>

<h4>Exercise 5: Generate a Design Checklist</h4> <ol> <li>Change Building Type to "Healthcare"</li> <li>Click "Set Project Context"</li> <li>Click "Generate Design Checklist"</li> <li>Note the healthcare-specific items added (ASHRAE 170, pressure relationships)</li> <li>Try with "Laboratory" and "Data Center" to see different checklists</li> </ol>

<h4>Exercise 6: Troubleshooting Scenario</h4> <ol> <li>Change mode to "Troubleshooter"</li> <li>Ask: "The conference rooms are always too warm in the afternoon. What should I check?"</li> <li>Review the diagnostic approach suggested</li> <li>Follow up: "Could it be related to the VAV boxes?"</li> </ol>

<h4>Sample Project Workflow:</h4> <pre class="bg-dark text-light p-3 rounded"><code>1. Set context: 25,000 SF medical office, Phoenix (2B), 150 occupants 2. Project Manager mode: "What HVAC system type should I use?" 3. Load Calculator mode: "Estimate my cooling and ventilation loads" 4. Equipment Selector mode: "Recommend packaged RTUs for this project" 5. Code Compliance mode: "What are the ASHRAE 170 requirements?" 6. Energy Optimizer mode: "How can I meet Title 24 efficiency requirements?" 7. Generate checklist and export conversation</code></pre>

<h4>Key Takeaways:</h4> <ul> <li>Set project context before detailed queries</li> <li>Use the appropriate mode for your task</li> <li>Ask specific questions for specific answers</li> <li>Always verify critical information</li> <li>Use Ollama for privacy-sensitive projects</li> <li>The AI accelerates research, but you make the engineering decisions</li> </ul>