Pillar 01 — AI

AI isn't magic. It's leverage — if you know where to point it.

Most small businesses are either ignoring AI completely or using it in ways that create liability without value. Both are expensive mistakes.

I started teaching AI to small business owners because the mainstream conversation was useless to them. Either it was enterprise case studies that don't translate, or it was breathless hype that skipped straight to “AI will replace your team.” Neither helped anyone actually do anything.

The reality is more interesting and more immediately actionable. AI tools — specifically large language models — are very good at a narrow set of things that happen to be bottlenecks in almost every small business: drafting, summarizing, classifying, and generating first versions of things that humans then improve.

In every session we focus on workflows that you can implement that week. Not pilots. Not proofs of concept. Production workflows that save real hours and create real output. We also spend time on what AI gets wrong, because building on shaky foundations is worse than not building at all.

By the end of a session, every person in the room leaves with at least one automation running and a policy that protects them from the mistakes the previous cohort made.

What we cover

Six concepts. All of them practical.

[A]

What LLMs actually are

Language models predict the next token. That's it. Understanding this single fact tells you everything about where they succeed and where they confidently hallucinate. They are not search engines. They are not databases. They are pattern-completion engines trained on text.

[B]

The three high-ROI workflows

Customer communication drafting, internal knowledge retrieval, and first-draft content production. These three cover the vast majority of time savings available to a small business in the next 90 days. Everything else is optimization.

[C]

Prompt engineering that actually works

Role, context, constraint, format. That's the whole framework. You don't need a course. You need to stop treating AI like Google and start treating it like a capable contractor who has never worked at your company before.

[D]

Automation pipelines

Connecting AI to your real workflows via tools like Make, Zapier, or n8n. We walk through a live example of an intake-to-response automation that a service business can have running the same week.

[E]

What you're leaking

Pasting customer PII into ChatGPT. Uploading contract language into free tools. Using personal accounts for business AI usage. These are real incidents happening inside real businesses right now, and the fix takes 20 minutes.

[F]

Building your AI policy

A one-page policy that tells your team what they can use, on what data, for what purposes. Not a legal document — an operational guardrail. We build one together during the session.

What trips people up

The four mistakes that make AI a cost center instead of a force multiplier.

The mistake

Using AI for everything at once

The fix

Pick one workflow. Nail it. Then expand. Broad rollouts fail because nobody owns anything.

The mistake

Asking AI to think for you

The fix

Use AI to produce volume, speed, and first drafts. Your judgment on strategy, customer relationships, and brand voice stays human.

The mistake

Trusting outputs without verification

The fix

Hallucination is a feature, not a bug — it's how the model fills gaps. Any factual claim needs a second source.

The mistake

Ignoring the data you're feeding it

The fix

What goes into a model prompt can be logged, stored, and used for training. Know your tool's data policy before you paste anything sensitive.

The stack

What tools, and when to use which.

ChatGPT / Claude

Drafting, ideation, summarization, Q&A on documents

Use for anything where iteration and conversation matter. Claude tends to follow nuanced instructions more reliably. ChatGPT has broader plugin integrations.

GitHub Copilot / Cursor

Code generation and review — relevant even for non-developers

If you have internal tools, scripts, or spreadsheet logic, AI code assistance cuts build time by 30–60%. You don't need to be a developer to benefit.

Make / Zapier / n8n

Connecting AI outputs to your existing tools

This is where AI becomes an automated employee. Build once, run continuously. We wire up a working example in every session.

Next session

We cover AI in every training. Come build something real.

Free, in-person, monthly across Washington — with an online option. Bring a real workflow you want to automate. Leave with it running.