← All Learning Paths
Intermediate Path

Working With AI Daily

You've tried AI a few times. Now make it a real part of your workflow — advanced prompting, document automation, image generation, data analysis, and the tools that multiply your output.

6 lessons ~60 min total Assumes basic AI familiarity
01

Advanced Prompting: Context, Personas, and Chains

12 min

The CRISP Framework

The difference between amateur and professional AI use comes down to five elements:

  • C — Context: Who are you? What's the situation? "I'm a QAQC manager on a 50MW datacenter project in Phase 2 commissioning"
  • R — Role: Who should the AI be? "Act as a senior mechanical engineer reviewing commissioning data"
  • I — Instructions: What exactly do you want? "Compare these two cooling system designs and recommend one"
  • S — Specifics: Format, length, tone, audience. "Create a comparison table, 1 page max, for a non-technical owner's rep"
  • P — Parameters: Constraints. "Use only data I provide — don't make assumptions. Reference ASHRAE 90.4 where applicable"

Persona Prompting

Telling AI to "act as" a specific role dramatically changes output quality. "Act as a construction attorney reviewing this contract clause" produces legalese-aware analysis. "Act as a project controls engineer" produces schedule-focused responses. The AI adjusts vocabulary, priorities, and depth to match the role.

Chain Prompting — Multi-Step Tasks

Break complex work into sequential steps within one conversation:

  • "First, read this spec section and list every testable requirement"
  • "Now organize those requirements by system (mechanical, electrical, plumbing)"
  • "Create a field verification checklist from that organized list, with pass/fail columns"
  • "Finally, add the applicable code references next to each item"

Each step builds on the previous one. The AI holds context across the entire chain. This is how you produce complex deliverables that would take hours — in minutes.

Try This

Take a task you do regularly — writing a report, reviewing a document, preparing for a meeting. Write a CRISP prompt for it. Then send it to Claude or ChatGPT. Compare the output to what you'd normally spend 30+ minutes producing.

02

AI for Documents, Spreadsheets, and Presentations

10 min

Document Processing

Upload or paste any document into Claude or ChatGPT and transform it:

  • Spec → Checklist: "Convert this specification section into a field verification checklist with pass/fail criteria"
  • Contract → Summary: "Summarize the key obligations, deadlines, and liability clauses in this subcontract — highlight anything unusual"
  • Meeting notes → Action items: "Extract all action items, who's responsible, and due dates from these meeting notes"
  • Long report → Executive brief: "Condense this 30-page report into a 2-page executive summary for the project owner"

Spreadsheet Power

AI can generate formulas, analyze data patterns, clean messy data, and create structured outputs. Paste your data (or describe it) and ask:

  • "Write an Excel formula that flags any value in column D that exceeds the tolerance in column E"
  • "I have 200 rows of inspection data. Summarize the deficiency trends by trade and building"
  • "Create a pivot table structure that shows deficiencies by system, location, and status"

Presentations in Seconds

Gamma (gamma.app) is the standout tool here. Describe your presentation: "Create a 10-slide deck on direct-to-chip cooling systems for a client audience — include market size, key players, technology comparison, and implementation considerations." You get a polished, designed deck in about 60 seconds. Edit, refine, present.

The workflow that saves hours: Use Claude or ChatGPT to generate the content and structure. Copy it into Gamma for the visual design. Review and refine. What used to take an afternoon takes 20 minutes.
03

Image Generation and Editing with AI

8 min

What AI Image Tools Can Do

Describe an image in words, get a high-quality visual in seconds. This is transformative for presentations, proposals, marketing, and concept visualization — no graphic design skills required.

The Tools

  • DALL-E (in ChatGPT): Built into ChatGPT Plus. Good for quick images, illustrations, and concepts. "A modern datacenter facility at sunset, aerial view, photorealistic"
  • Midjourney: The highest-quality option. Produces stunning, professional visuals. Accessed through Discord. Best for marketing materials and presentations.
  • Adobe Firefly: Integrates with Adobe Creative Suite. Best if you already use Photoshop, Illustrator, or other Adobe tools.

Practical Uses

  • Proposals: Generate concept renderings of facilities, equipment layouts, or project visualizations
  • Presentations: Create custom hero images, diagrams, and illustrations instead of using stock photos
  • Marketing: Social media graphics, website visuals, brochure imagery — all custom, all on-brand
  • Training: Visual aids for safety training, equipment identification, process flows
Prompting for images: Be specific about style, perspective, lighting, and content. "A modern datacenter server room, rows of racks with blue LED lighting, wide angle, clean and minimal, professional photograph style" works much better than "datacenter."
04

AI for Data Analysis and Decision-Making

10 min

Turn Raw Data Into Insights

Upload a spreadsheet of inspection results, test data, or project metrics. Ask AI to find what matters:

  • "Which trade has the highest deficiency rate per inspection?"
  • "Are there any trends in these commissioning test failures — by system, building, or time period?"
  • "Compare our actual progress against the baseline schedule and flag areas more than 5% behind"
  • "Identify statistical outliers in this temperature sensor data that might indicate equipment issues"

Decision Support (Not Decision Making)

AI is excellent at laying out options with pros, cons, and trade-offs — but the decision is always yours. Use it to structure your thinking:

  • "I'm choosing between two cooling system vendors. Here are the specs and pricing. Create a weighted comparison matrix"
  • "We're behind schedule on the electrical rough-in. What are our options for recovery, and what are the cost and risk implications of each?"
Try This

Take any spreadsheet you work with regularly — inspection logs, test results, cost tracking. Paste a sample into Claude and ask: "Analyze this data. What patterns do you see? What should I be paying attention to?" The insights often surface things you'd miss scanning rows manually.

05

Automating Repetitive Work

10 min

The Automation Mindset

Every time you think "I do this exact same thing every day/week," that's an automation candidate. The goal isn't to automate everything — it's to automate the parts that don't need your brain, so your brain is free for the parts that do.

Level 1: Templates + AI

Create a prompt template you reuse. Save it in a note or document. Every time you need a daily report, paste the template, fill in today's details, send to AI. You've just turned a 30-minute task into a 3-minute task — no technical setup required.

Level 2: Claude Projects / Custom GPTs

Upload your company templates, SOPs, and reference documents into a Claude Project or Custom GPT. Now every conversation in that workspace has your context loaded. You don't re-explain the project every time — the AI already knows your format, your standards, and your team's terminology.

Level 3: Zapier / Make

Connect your apps with AI in the middle. Real examples running right now on construction projects:

  • Email arrives with "RFI" in subject → AI extracts the question → Logs it in your tracker → Drafts a response → Sends to responsible engineer for review
  • 5 PM daily → Pull today's entries from project management tool → AI generates daily report → Emails to PM and owner's rep
  • Photo uploaded to project folder → AI analyzes for safety issues → Flags anything concerning → Notifies safety manager
Start with Level 1. A saved prompt template costs nothing, takes 5 minutes to set up, and starts saving time immediately. You can always level up later.
06

Choosing the Right AI Tool for the Job

10 min

The Decision Matrix

Don't use one tool for everything. Each AI tool has strengths:

  • Deep analysis, long documents, careful writing → Claude
  • General tasks, quick answers, image generation → ChatGPT
  • Google Workspace integration → Gemini
  • Research with citations → Perplexity
  • Presentations → Gamma
  • High-quality images → Midjourney
  • Workflow automation → Zapier + AI steps
  • Meeting transcription → Otter.ai
  • Coding and technical builds → Claude Code / Cursor

Free vs. Paid — Is It Worth It?

Free tiers are genuinely useful. Paid tiers ($20/month for Claude Pro or ChatGPT Plus) give you faster responses, longer conversations, more features, and higher usage limits. If you're using AI more than a few times a week, the paid tier pays for itself in the first hour of time saved. That's not marketing — it's math.

Building Your AI Stack

Most productive professionals in 2026 use 2-3 AI tools regularly:

  • Primary: Claude or ChatGPT for daily work (pick one as your main tool)
  • Specialized: One tool for your biggest time-sink (presentations, automation, or images)
  • Optional: A meeting tool (Otter.ai) if you're in meetings daily
The productivity stack that pays for itself: Claude Pro ($20/mo) + Gamma ($10/mo) + Zapier ($20/mo) = $50/month. If this saves you 5 hours/week — and it will — that's $50 buying back 20+ hours per month. No tool investment in history has had that ROI.

Ready to Go Pro?

You're using AI effectively in your daily work. Now learn to build it into your business — agents, custom workflows, and automation at scale.

Continue to Professional Path →