AI for CRE and A&D

Prompting and Tools for Projects and Personal Productivity

60-Minute Interactive Workshop

by Joe Scanlin

Warm-Up: Why This Feels Different

The Autonomous Driving Analogy

Autonomous driving analogy for AI-assisted work

You're still driving. You're still making the important decisions.
You're just not exhausted by the micro-decisions anymore.

90% reduction in cognitive load = Energy preserved for strategic thinking

0-5 min

Kickoff and Outcomes

Why AI Now?

AI has moved from experimental to essential for project teams
Designers and client-facing roles gain 5-15 hours per week
Shift from "nice to have" to competitive advantage
Tools are production-ready, patterns are proven

Interactive Poll: Your Biggest Time Drains
                        Research & intel gathering
Proposal writing
Spec drafting
Meeting prep

                    

                        Email triage
RFI responses
Planning & coordination
Documentation

                    

By the end: You'll have prompts, patterns, and a daily routine that saves 5+ hours per week.

5-12 min

What to Use AI For (and Not For)

✓ Great Fits

Project work: Research, stakeholder mapping, pursuit strategy
Documents: Proposals, specs, RFI triage, meeting notes
Design prep: Charrette planning, discovery questions
Personal: Daily planning, calendar/email triage, summaries

⚠ Cautions

Confidentiality: Never paste client-specific data
IP: Don't share proprietary designs or methods
Compliance: Always verify code requirements
Citations: Require sources, check verifiability

Golden Rule

Separate client data from personal data. Anonymize project details before pasting.

Example: Instead of "The Apple headquarters renovation in Cupertino", use "A 500K sf corporate headquarters renovation in Northern California."

12-15 min

What's Actually Happening?

How Large Language Models Work (Simple Version)

Think of it Like This

An LLM is like an incredibly well-read assistant who's seen billions of examples of text and learned what words tend to follow other words in different contexts.

What It's Good At

Finding patterns in language
Completing sentences naturally
Matching style and tone
Following common structures

What It's Not

Not "thinking" or "understanding"
No actual knowledge or memory
Can't access real-time data
Doesn't fact-check itself

Key Takeaway: LLMs are prediction engines trained on patterns. That's why good prompts matter—you're giving it better context to make better predictions.

15-28 min

Prompting That Works

The CRAFT Framework

Every effective prompt includes Context, Role, Action, Format, and Tone

Example: Before CRAFT
                    "Tell me about this project."
                
After CRAFT
                    "Act as a project pursuit lead. Using the project details below, produce: 1) scope and drivers (3 bullets), 2) stakeholders and likely priorities (table format), 3) three insight-led openers for outreach. Keep it under 120 words per section. Professional but conversational tone."
                

12-25 min (cont.)

Proven Prompting Patterns

Pattern 1: Write Then Refine

Step 1: "Draft a 150-word project summary from these notes: [paste]" Step 2: "Now refine it for a client-facing email. Make it 75 words, add a clear next step, and use a warm professional tone."

Pattern 2: Three Options

"Give me three versions of this outreach email: 1) direct and brief, 2) relationship-focused with a project reference, 3) insight-led with an industry trend. 50-75 words each."

Pattern 3: Request Missing Info First

"Before drafting, ask me for any missing info you need about: target audience, constraints, desired outcome, timeline, and must-include points."

Pattern 4: Compare A vs B

"Compare these two HVAC systems using a scoring rubric: cost (0-10), efficiency (0-10), maintenance (0-10), lifecycle (0-10). Show scores in a table, then recommend one with rationale."

12-25 min (cont.)

Guardrails for Quality

Always Add These Constraints

Sources: "Include sources for all claims, or say 'unknown' if unavailable"
Audience: "Write for a [facility manager / C-suite exec / design team]"
Length: "Keep under [150 words / 3 bullets / 1 page]"
Structure: "Use [bullets / table / paragraph] format"

Personal Task Self-Check

For calendar planning and inbox responses, add this line:

"Before finalizing, check: Are there any conflicts? Did I account for travel time? Is the priority order realistic?"

Example with Guardrails

"Act as a sustainability consultant. Create a lifecycle cost comparison for two roof systems: TPO vs. green roof. Include material, install, maintenance (20 years), and replacement interval. Output as a table. Cite sources for cost data or note 'estimate.' Write for a facility director audience."

25-35 min

Project Research & Pursuit Strategy

Live Demo Walkthrough

Inputs You'll Use

Project name or address (anonymized if needed)
Firm websites and LinkedIn profiles
Public news articles or press releases
Award lists or similar case studies

Outputs You'll Get
                        One-paragraph project brief
Stakeholder map with priorities
Three insight-led openers

                        75-word outreach email
Timeline bullets
Risk flags

Demo Prompt:

"Act as a project pursuit lead. Using the project info below, produce: 1) scope and drivers (3 bullets), 2) stakeholders and likely priorities (table: name, role, goal, preferred evidence), 3) three insight-led openers for outreach, 4) a 75-word email, 5) timeline and risk bullets. Keep it under 120 words per section.

[Paste anonymized project details]"

25-35 min (cont.)

Personal Productivity Add

10-Minute Morning Routine

Transform morning chaos into a clear plan in just 10 minutes

35-45 min

Meetings, Charrettes, and Objections

Discovery Questions by Discipline

"Generate 5 discovery questions for a [owner / architect / GC / facilities manager] interview about [project type]. Focus on: priorities, constraints, past pain points, decision criteria, and timeline."

Meeting Notes → Action Plan

"From these meeting notes, create:
1) Recap paragraph (what was decided)
2) Decision criteria (what matters most)
3) Dated next actions with owners
4) Open risks or dependencies

[Paste raw notes or transcript]"

Handling Objections: Validate + Evidence + Close

Common objections: Budget, schedule, code/certification, maintenance cost, product availability

"Draft a response to this objection: [paste]. Use the pattern: 1) validate their concern, 2) provide evidence or alternatives, 3) close with a clear next step. Keep under 100 words."

35-45 min (cont.)

Personal Add: One-Click Follow-Up

The Problem

Meeting ends. You have notes. You need to send a recap, create tasks, and schedule follow-ups. It takes 20 minutes you don't have.

The Solution: Follow-Up Routine

"From these meeting notes, create:

1) Email recap for attendees (100 words, include decisions and next steps)
2) Task list with owners and due dates
3) Calendar items I should schedule

[Paste notes]"

Example: Meeting on HVAC System Selection

Input (raw notes)

"Discussed two HVAC options. Team leaning toward VRF for flexibility. Budget concern: $80k delta. FM worried about maintenance training. Need vendor quotes by Friday. Code review needed."

Output (structured)

Email: "We're moving forward with VRF system pending vendor quotes (due Fri). Budget delta $80k—exploring value engineering. FM training plan needed."

Tasks: Request quotes (Sarah, Fri), code review (Tom, Wed), training plan (Lisa, Mon)

45-53 min

Discipline-Specific Plays

Architect and Designer

Tech sheet → Client one-pager: "Convert this technical sheet into a client-facing summary: Problem, Why this option, Performance/code notes, Install overview, Maintenance, Next step. Bullets only."
Spec language: "Draft spec language for [product/system] using MasterSpec format. Include performance criteria, installation requirements, and acceptance testing."
Value engineering: "List three VE alternatives for [space/product]. Compare: cost, schedule, performance, maintenance. End with recommendation and risks."

Owner and Facilities Manager

Lifecycle cost: "Build a TCO comparison for [two options]. Include material, install, maintenance, replacement interval (20 years), and rationale. Table format."
Risk register: "From this project scope, create a risk register: risk description, likelihood (L/M/H), impact (L/M/H), mitigation strategy, owner."
Change log: "Summarize these change orders into a table: ID, description, cost impact, schedule impact, approval status, reason."

Contractor and Rep

Bid summary: "Summarize these bids into a comparison table: bidder, total cost, key inclusions/exclusions, timeline, payment terms, concerns."
RFI triage: "Summarize these RFIs: ID, topic, owner, due date, dependencies. Flag code or certification risks."
Install sequencing: "Create install sequence notes for [system]. Include: prerequisites, duration, crew size, coordination with other trades, inspection points."

45-53 min (cont.)

Personal Productivity Templates

Weekly Review

"Review my week:
• Completed: [list of completed tasks]
• Carried over: [incomplete items]
• Meetings: [hours in meetings this week]

Output: 1) Three wins to celebrate, 2) Two process improvements, 3) Next week's top three priorities."

Timeboxing Assistant

"I have [list tasks]. My available time blocks today: [list from calendar]. Assign each task to a block based on: energy level needed, dependencies, and deadline urgency. Flag if I'm overcommitted."

Inbox Zero Lite

"From these subject lines and snippets [paste], group by action needed: 1) reply now (draft 3 short responses), 2) schedule for later (suggest times), 3) delegate (note to whom), 4) archive. Propose a 25-minute batch to clear group 1."

53-58 min

Hands-On Sprint

Apply What You've Learned

Exercise: Build Your Own Prompts

Use a real project or task from your work. Create:

A 120-word personalized outreach email for a project
Three discovery questions tailored to your stakeholder
A compelling subject line (under 60 characters)

Apply the CRAFT Framework

Context: Include project details, your role, the recipient's role

Role: "Act as a [your role] reaching out to a [their role]"

Action: "Draft outreach + discovery questions + subject line"

Format: "120 words for email, 3 bullets for questions"

Tone: "Professional but warm, insight-led"

Remember: Anonymize client details. Use "a 200K sf office building" instead of specific names or addresses.

Pair and Share

Share your prompt and output with a partner. Give feedback on: clarity, specificity, guardrails, and whether it solves the right problem.

53-58 min (cont.)

Personal Sprint: Morning Planner

Build Your Two-Line Morning Prompt

Create a reusable prompt you can use tomorrow morning (and every morning after).

Template

"Create a 10-minute plan for today using my calendar blocks [paste], top three priorities [list], and inbox count [number]. Output: 1) three tasks with timeboxes, 2) one email batch window, 3) one short break. Check for conflicts and travel time."

Customize It

Add constraints that match your work style:

"I work best on deep tasks in the morning, meetings in the afternoon"
"Block 30 minutes before any client call for prep"
"I need a walk break by 2pm or I lose focus"
"Batch all email/admin work, don't scatter it"

Pro tip: Save this prompt in a note or doc titled "Morning Routine." Just copy, paste your calendar, and run it daily.

58-60 min

The 5-Point AI Practitioner Checklist

1. Always include Context, Role, Action, Format, Tone
This is your baseline. Every prompt should have all five.

2. Ask the model to list missing info before drafting
"Before drafting, ask me for anything you need to know."

3. Constrain output length, structure, and audience
"Under 120 words, bullet format, for a facility director."

4. Require sources or "unknown"
"Cite sources for all claims, or note 'unknown' if unavailable."

5. Iterate to a score target for your persona
"Rate this on clarity (0-10), professionalism (0-10), actionability (0-10). If under 8 on any, revise."

Print this checklist. Put it next to your monitor. Reference it every time you prompt.

58-60 min

Your Copy-Paste Prompt Library

Ready-to-use templates for common tasks

The next slides contain 10 prompts you can copy and use immediately. Just replace the {bracketed} sections with your specific details.

                How to Use These Prompts
                Copy the entire prompt
Replace {bracketed text} with your details
Paste into ChatGPT, Claude, or your preferred tool
Review output, iterate if needed
Save successful prompts for reuse

            

Remember: Always anonymize client-specific information before pasting into any AI tool.

Prompt Library: Project Work

Project Brief Builder

"Act as a {project pursuit lead}. Using {project name or URL}, produce: 1) scope and drivers, 2) stakeholders and likely priorities, 3) three insight led openers. Bullets only, no more than 120 words per section."

Stakeholder Map

"Given {firm names and roles}, infer decision influence for {project type}. Output a table: stakeholder, goal, risk, preferred evidence."

Proposal Executive Summary

"Turn these notes into a 150 word executive summary with three bullets on outcomes, two bullets on risks and mitigations, and a clear next step."

Meeting Recap to Action Plan

"From these notes, create: 1) recap paragraph, 2) decision criteria bullets, 3) dated next actions with owners, 4) open risks."

Prompt Library: Technical Work

Spec Assist

"Convert this {technical sheet text} into a client facing one pager: Problem, Why this option, Performance and code notes, Install overview, Maintenance, Next step. Bullets only."

Value Engineering Alternatives

"List three value engineering paths for {space and product}. Compare cost, schedule, performance, maintenance. End with a recommendation and risks."

RFI Triage

"Summarize these RFIs into a table: ID, topic, owner, due date, dependencies. Flag any code or certification risks."

Lifecycle Cost Snapshot

"Build a simple TCO view for {two options}. Include material, install, maintenance, replacement interval, and a one line rationale."

Prompt Library: Personal Productivity

Personal Planner (Morning Routine)

"Create a 10 minute plan for today using my calendar blocks {paste}, top three priorities {list}, and inbox count {number}. Output: 1) three tasks with timeboxes, 2) one email batch window, 3) one short break."

Inbox Triage Helper

"From these subject lines and snippets {paste}, group by action needed, draft three short replies, and propose a 25 minute batch to send."

Weekly Review

"Review my week: completed {list}, carried over {list}, meeting hours {number}. Output: 1) three wins, 2) two process improvements, 3) next week's top three priorities."

Timeboxing Assistant

"I have {list tasks}. Available time blocks: {list from calendar}. Assign each task based on energy level, dependencies, and urgency. Flag if I'm overcommitted."

58-60 min

Tool Starter Pack & Next Steps

Recommended Tools

Core AI Tools

ChatGPT or Claude for drafting and reasoning
Perplexity for research with citations

Project Intelligence

Firm and project websites
Award databases
News and press releases

Specs & Technical

MasterSpec or SpecLink
ARCAT, BIMobject, BIMsmith for cut sheets

Meetings & Automation

Otter, Fathom, Gong for notes
Zapier or Make for workflows

                Your 10-Minute Daily Routine
                Morning (5 min): Run your planner prompt with calendar + priorities
Midday (3 min): Triage inbox, draft quick responses
Evening (2 min): Review action items, prep tomorrow's top 3

            

You now have the prompts, patterns, and practices to save 5+ hours per week.

Start with one prompt tomorrow. Build from there.

→ Continue to Appendix for deeper insights and tools

Appendix

Deeper Insights, Tools, and Advanced Strategies

• Before/After Time Audit

• Common Pitfalls & Fixes

• Tool Selection Matrix

• 4-Week Adoption Roadmap

• Client-Facing Output Examples

• Integration with Your Tech Stack

• The 2-Minute Wins

• Error Recovery Patterns

• Team Adoption Playbook

Appendix A1: Your Week Before & After AI

The Real Time Savings (Not Hypothetical)

Without AI

30 hrs on micro-tasks
10 hrs on strategic work
Constant context switching
Exhausted by Wednesday

With AI

8 hrs on micro-tasks (saved 22 hrs!)
32 hrs on strategic work
Deep focus preserved
Ahead of schedule

ROI Calculation: If your billable rate is $150/hr, saving 22 hours/week = $3,300/week or $165K/year in recovered time

Appendix A2: Common Pitfalls & How to Fix Them

Why your prompts aren't working (and what to do)

Pitfall #1: Too Vague

❌ Doesn't Work

"Help me with this project."

✓ Works

"Act as a project lead. Create a 3-bullet scope summary for a 200K sf office renovation with a 12-month schedule. Include key constraints."

Fix: Add Context, Role, Action, Format, Tone (CRAFT)

Pitfall #2: No Constraints

❌ Doesn't Work

"Write an email about the meeting."

Result: 300-word essay, wrong tone, generic

✓ Works

"Write a 75-word follow-up email for a C-suite audience. Professional but warm tone. Include: decision made, next action, date."

Result: Concise, appropriate, actionable

Fix: Specify length, audience, structure, tone

Pitfall #3: Accepting First Output

Common mistake: "The AI gave me something mediocre, so AI doesn't work."

Reality: AI output is a starting point. Use "write then refine" pattern:

Get initial draft
Review and identify what's off
Prompt: "Make it more [specific fix]"
Iterate 2-3 times to perfection

Pitfall #4: No Quality Check

Always verify before using:

Are there factual claims? → Ask for sources
Is it client-facing? → Check tone and accuracy
Are there numbers? → Verify calculations
Code compliance mentioned? → Double-check with official sources

Remember: AI assists, you approve.

Appendix A3: Which AI Tool for Which Task?

Decision Matrix for CRE & A&D Work

Task Type	Best Tool	Why This Tool	Time Saved
Quick drafts (email, summaries)	ChatGPT or Claude	Fast, conversational, handles ambiguity well	80%
Research with sources (codes, case studies)	Perplexity	Includes citations, web search integrated	90%
Long documents (proposals, specs)	Claude	200K token context, handles large inputs	75%
Meeting transcripts (notes → action items)	Otter.ai + ChatGPT	Otter captures audio, ChatGPT structures output	85%
Technical specs (product data)	ChatGPT + ARCAT	ARCAT for data, ChatGPT to format client-facing	70%
Image generation (concept boards)	DALL-E or Midjourney	Purpose-built for visual content	Variable
Workflow automation (email → CRM)	Zapier or Make	Connects AI outputs to your existing tools	95%

                Quick Selection Guide
                Need it fast + conversational? → ChatGPT
Need citations/sources? → Perplexity
Working with long documents? → Claude
Want to automate repetitive tasks? → Zapier/Make
Need meeting transcription? → Otter.ai or Fathom

            

Pro Tip: Most professionals use 2-3 tools regularly. Start with ChatGPT or Claude for 90% of tasks, add Perplexity for research. Don't overcomplicate.

Appendix A4: Your 4-Week Adoption Roadmap

Slow and steady wins the race

                Why This Approach Works
                Week 1: Build one habit. Don't overwhelm yourself. Morning planner = foundation.
Week 2: Layer on two more. You're comfortable with prompting now.
Week 4: Full integration. AI is your daily co-pilot, not a novelty.
Beyond: Customize and optimize for your specific workflow.

            

Common mistake: Trying to use AI for everything on Day 1.
Result: Overwhelm → giving up by Week 2.

Better approach: Start small, prove value to yourself, build momentum.

Set Yourself Up for Success

Add calendar reminders: "Week 2: Try meeting recap prompt"

Save your successful prompts in a note/doc for reuse

Track time saved (even roughly) to stay motivated

Share wins with a colleague or accountability partner

Appendix A5: Client-Facing Output Examples

Real AI outputs you can trust

Example 1: Stakeholder Map (Generated in 90 seconds)

Stakeholder	Role	Priority	Evidence They Value
Sarah Chen	VP Facilities	Lifecycle cost, maintenance	TCO analysis, service contracts
Marcus Williams	CFO	Budget, ROI timeline	Financial models, payback period
Linda Park	Head of Design	Aesthetics, brand alignment	Renderings, material samples

Prompt used: "Given [3 names and roles], infer decision influence for [200K sf renovation]. Output a table: stakeholder, priority, evidence."

Example 2: 75-Word Outreach Email (Generated in 45 seconds)

Subject: Jefferson Tower HVAC Upgrade – Quick Question

Hi Sarah,

I saw your team is evaluating HVAC options for Jefferson Tower. We've worked on three similar retrofits in the past year, each achieving 30%+ energy savings without major tenant disruption.

Would a 15-minute call next week make sense to share what worked? Happy to adjust to your schedule.

Best,
[Your Name]

Prompt used: "Draft a 75-word outreach for a facilities VP about an HVAC project. Insight-led opener, warm professional tone, clear next step."

Example 3: Meeting Recap (Generated in 2 minutes)

HVAC Selection Meeting – Jan 15, 2025
Decision Made:
Move forward with VRF system for flexibility. Budget delta ($80K) approved pending value engineering review.
Next Actions:
Sarah: Request vendor quotes by Friday (1/19)
Tom: Code compliance review by Wednesday (1/17)
Lisa: Draft FM training plan by Monday (1/22)
Open Risks:
Lead time for equipment could affect Q2 timeline. Vendor quotes will clarify.

Prompt used: "From these notes, create: 1) decision paragraph, 2) next actions with owners/dates, 3) open risks."

Notice the quality level: These outputs are ready to send (with minor review).
Not "AI slop" — professional, specific, actionable.

This is what's possible when you use CRAFT prompting.

Appendix A6: Integration with Your Tech Stack

How AI fits into your existing workflow

The Integration Pattern

Your Tools
Outlook, Gmail,
Teams, Slack
→
AI Layer
ChatGPT, Claude,
Perplexity
→
Your Output
Docs, CRM,
PM Tools

Common Integration Workflows

Email → AI → Task Manager

Copy email thread
Paste into ChatGPT with prompt: "Extract action items with owners and due dates"
Copy output to Asana/Monday/Trello

Advanced: Use Zapier to automate steps 2-3

Meeting → AI → Client Follow-Up

Record with Otter.ai or Fathom
Export transcript to ChatGPT
Prompt: "Create recap email"
Review, refine, send from Outlook

Time saved: 20 min → 3 min

Research → AI → Proposal

Gather project info (website, news)
Paste into Perplexity for research
Feed output to Claude: "Draft proposal executive summary"
Export to Word, format, finalize

Time saved: 2 hrs → 20 min

Calendar → AI → Daily Plan

Copy calendar for today
Paste into ChatGPT with morning planner prompt
Get timeboxed plan with tasks
Set reminders or block calendar

Time saved: 15 min → 2 min

Automation Options (Advanced)

Once you're comfortable with manual AI workflows, consider automating repetitive tasks:

Zapier: Connect Gmail → ChatGPT API → Google Sheets for automatic email triage
Make (formerly Integromat): More complex workflows, visual automation builder
Native integrations: Otter.ai → Slack, Fathom → CRM, ChatGPT → Notion

Start here: Master manual prompting first (Weeks 1-4). Then add automation where it makes sense.

Important: You don't need automation to save 10+ hours/week. Manual copy-paste workflows are fast and flexible. Automation is a bonus, not a requirement.

Appendix A7: The 2-Minute Wins

Quickest impact prompts you can try right now

Perfect for: On your phone during a break, between meetings, waiting for a call to start
Each takes ~2 minutes to execute and delivers immediate value

Win #1: Smart Subject Lines (30 seconds)

"Write 3 subject line options for this email: [paste draft]. Make them under 60 characters, action-oriented, and professional."

Use case: Boost open rates on client emails

Win #2: Meeting Agenda Generator (90 seconds)

"Create a 30-minute meeting agenda for [topic]. Include: objective, 3-4 discussion points with time allocations, decision to be made, and next steps section."

Use case: Run more focused meetings with clear outcomes

Win #3: Instant Meeting Prep (2 minutes)

"I'm meeting with [role/company] about [topic] in 10 minutes. Give me: 3 smart opening questions, 2 potential concerns they might have, and 1 concrete next step to propose."

Use case: Walk into meetings confident and prepared

Win #4: Email Tone Checker (45 seconds)

"Rate this email on: clarity (0-10), professionalism (0-10), warmth (0-10). If any score is below 8, suggest specific fixes. [paste email]"

Use case: Catch tone problems before you hit send

Win #5: LinkedIn Connection Note (60 seconds)

"Write a 50-word LinkedIn connection request to [name/role]. Reference [mutual connection or shared interest]. Professional but personable."

Use case: Network more effectively without sounding generic

🎯 Challenge: Try one of these TODAY. Pick the lowest-friction option and just do it.
Once you experience the speed, you'll be hooked.

Appendix A8: Error Recovery Patterns

What to do when AI gets it wrong

Reality check: AI will make mistakes. It will hallucinate facts, miss nuance, and occasionally produce garbage. The difference between novices and pros isn't avoiding errors—it's recovering from them quickly.

Pattern #1: The Factual Hallucination

❌ The Problem

AI confidently states: "Building code requires X" but you know that's wrong.

✓ Recovery Prompt

"I need to verify this claim. What's your source for [statement]? If you're uncertain, say 'I should verify this' instead of guessing."

Lesson: Always ask for sources on factual claims. AI often admits uncertainty when pressed.

Pattern #2: Wrong Tone / Style

❌ The Problem

Output is too formal, too casual, or uses jargon your audience won't understand.

✓ Recovery Prompt

"Rewrite this for a [specific audience]. Make it more [conversational / technical / executive-friendly]. Avoid [jargon / fluff / passive voice]."

Lesson: Be specific about what's wrong. "Make it better" doesn't work. "Make it 30% more conversational" does.

Pattern #3: Missing Context

❌ The Problem

AI gives generic advice that doesn't account for your specific constraints.

✓ Recovery Prompt

"Revise considering these constraints: [budget / timeline / location / regulations]. Prioritize [specific criteria]."

Lesson: AI can't read your mind. If output is generic, you didn't provide enough context upfront.

Pattern #4: Structural Problems

Problem: Output is organized wrong—bullets should be a table, paragraph should be bullets, etc.

Recovery Prompt:

"Reformat this as a [table / bulleted list / numbered steps / paragraph]. Columns should be: [specify]. Keep content identical, just change structure."

Lesson: Specify format upfront in your CRAFT prompt to avoid this.

Pattern #5: Complete Garbage Output

When to start over:

Output is 80%+ wrong or irrelevant
You've iterated 3+ times and it's still not close
The AI fundamentally misunderstood your request

Solution: Start a new conversation. Sometimes the thread has too much confusing context. Rewrite your prompt with CRAFT, paste into fresh chat, and try again.

Your Error Recovery Checklist

Factual claim? → Ask for sources

Wrong tone? → Specify exact changes needed

Too generic? → Add missing constraints

Wrong format? → Reformat, keep content

Total garbage? → Fresh conversation, better prompt

Appendix A9: Team Adoption Playbook

How to bring your team along (without forcing it)

The challenge: You're excited about AI, but your team is skeptical, overwhelmed, or worried about job security. How do you create momentum without mandating adoption?

Step 1: Lead with Your Own Results (Week 1-2)

What to Do

Use AI for your own work
Track time saved (roughly)
Share one win casually: "I used AI to draft that proposal in 20 minutes"
Don't evangelize—just demo results

Why This Works

Shows > tells
Reduces skepticism with proof
Makes it safe to ask questions
Creates curiosity, not pressure

Step 2: Offer to Help One Person (Week 3)

Example conversation:

"Hey Sarah, I know you're swamped with RFI responses. Want me to show you a 5-minute AI trick that could cut that time in half? No pressure—just thought it might help."

Key: Offer help, don't mandate adoption. Find their pain point (meetings, emails, proposals) and solve it together.

Step 3: Share a Starter Prompt Library (Week 4)

Create a shared doc with 5-7 prompts:

Morning planner prompt (personalizable)
Meeting recap → action items
Email triage batch
Proposal executive summary
Quick subject line generator

Label each: "Saves ~X minutes" and "Best for [role/task]". Make it copy-paste ready.

Step 4: Address Concerns Head-On

Common Objections

"AI will replace my job"
"I don't have time to learn this"
"AI makes mistakes"
"Clients will know it's AI"

Your Responses

"AI assists, it doesn't replace judgment"
"Start with one 5-min prompt per day"
"That's why we review before sending"
"With CRAFT prompting, quality is professional"

Step 5: Propose a Team Pilot (Month 2)

Pitch to leadership:

"I've been using AI for [task types] and saved ~[X] hours/week. I'd like to run a 4-week pilot with [3-5 team members] to test broader adoption. We'll track time saved, document best prompts, and report results. Low risk, high upside."

Include: Clear success metrics (hours saved, tasks accelerated), opt-in only, share learnings at the end.

Team Adoption Do's and Don'ts

✓ DO

Lead by example first
Share wins, not hype
Offer help, not mandates
Start with pain points
Track and report results

✗ DON'T

Force adoption top-down
Overpromise ("AI solves everything!")
Skip training/support
Ignore legitimate concerns
Expect instant buy-in

Bottom line: Adoption spreads organically when people see results, not when they're told to change.
Your job is to make the first step easy, safe, and valuable.