Lesson #2
Working With AI: Prompts, Data & Guardrails
- 1
Prompt
Fundamentals“PROMPT”
1/5 · EasyThe instructions or question you give an AI so it knows what you want.
Think of it like… Telling a new employee exactly what job to do.
Construction: 'Write a friendly estimate email for a 12x14 deck rebuild.'
Business: 'Summarize this contract in five bullet points a customer can understand.'
Why this matters: The quality of an AI's answer depends mostly on the quality of your prompt.
Open full entry → - 2
Prompt Engineering
Fundamentals“PROMPT en-jih-NEER-ing”
2/5 · Getting thereThe skill of writing instructions that get better answers from AI.
Think of it like… Learning how to ask a contractor for a bid so you get the exact numbers you need.
Construction: Telling the AI your region, materials, and labor rate before asking for an estimate.
Business: Giving the AI your brand voice and audience before asking for marketing copy.
Why this matters: Small changes in wording can turn a useless answer into a usable one — no coding required.
Open full entry → - 3
Training Data
Data“TRAY-ning DAY-ta”
2/5 · Getting thereThe information an AI studied in order to learn.
Think of it like… The textbooks and years of experience behind an expert's advice.
Construction: An estimating AI trained on thousands of past job invoices and material costs.
Business: A support AI trained on five years of customer service tickets.
Why this matters: If the training data is narrow or biased, the answers will be too — good data is the foundation.
Open full entry → - 4
Fine-Tuning
Models“FYNE TOON-ing”
3/5 · TrickierExtra training that teaches a general AI the specifics of your work.
Think of it like… Hiring an experienced carpenter, then teaching them how your company does things.
Construction: Training a general AI on your own bids so its estimates match your pricing.
Business: Teaching an AI your company's policies so replies sound like your team.
Why this matters: Fine-tuning turns generic AI into an AI that actually understands your business.
Open full entry → - 5
Hallucination
Fundamentals“huh-LOO-sih-NAY-shun”
2/5 · Getting thereWhen an AI confidently gives an answer that is simply made up.
Think of it like… A confident coworker who'd rather guess than admit they don't know.
Construction: An AI citing a building code section number that doesn't exist.
Business: An AI quoting a refund policy your company never had.
Why this matters: AI can be wrong while sounding certain — always verify facts that carry money or legal risk.
Open full entry → - 6
Context Window
Models“KON-text WIN-doh”
3/5 · TrickierHow much information an AI can hold in mind at one time.
Think of it like… A desk that only fits so many papers before something has to come off.
Construction: Pasting a 200-page spec book and the AI only using the part that fits.
Business: An AI summarizing a long contract but missing details from page one.
Why this matters: Knowing the limit explains why AI 'forgets' — and why feeding it the right pieces matters.
Open full entry → - 7
Retrieval-Augmented Generation (RAG)
Data“RAG”
4/5 · AdvancedA method where the AI looks up your real documents before answering, instead of guessing.
Think of it like… An employee who checks the filing cabinet before answering a customer.
Construction: An AI pulling your real permit records before answering a code question.
Business: An AI answering HR questions straight from your employee handbook.
Why this matters: RAG is the main way businesses stop hallucinations and make AI answer from their own truth.
Open full entry → - 8
API
Infrastructure“A-P-I”
3/5 · TrickierA doorway that lets two pieces of software talk to each other.
Think of it like… A restaurant server carrying orders between you and the kitchen.
Construction: Your scheduling software sending job details straight into your accounting software.
Business: A store's website checking inventory in the warehouse system in real time.
Why this matters: APIs are how AI plugs into the tools you already use instead of living on an island.
Open full entry → - 9
Automation
Infrastructure“aw-toh-MAY-shun”
1/5 · EasySetting up work so it happens by itself, without someone doing it each time.
Think of it like… A sprinkler on a timer instead of you dragging the hose out every evening.
Construction: Every finished job automatically triggering an invoice and a review request.
Business: New leads automatically added to the CRM and assigned to a rep.
Why this matters: Automation is where AI saves real hours — repetitive work stops needing a human.
Open full entry → - 10
Guardrails
Fundamentals“GARD-rayls”
2/5 · Getting thereThe rules and limits you set so an AI stays safe and on task.
Think of it like… The railing on a staircase — it doesn't slow you down, it keeps you from falling.
Construction: An AI that can draft an estimate but can never send it without your approval.
Business: A support AI that must hand off to a human for any refund over $200.
Why this matters: Guardrails are what make it safe to let AI act — they turn risk into trust.
Open full entry →