we're back.
AI year three.
Year 1 we learned what AI is. Year 2 we learned how to use it. This year — we put it to work. Here's your cheat sheet for everything we covered. Save it. Share it. Come back to it.
the vibes you came for.
Nine little reminders to tape inside your skull. Pulled from the talk. Use liberally.
AI is a gym membership
You only get strong if you show up. Build AI intuition by using it — small tasks, daily.
Permission slip, not detention slip
Write an AI policy your team can actually say yes to. Encourage smart experimentation, not fear.
It hallucinates — so verify
Only ask about things you can fact-check. Ask it to cite sources. Human oversight = the power combo.
Don't just chat — build a system
Every good output becomes a template. Save it. Reuse it. Get faster every time.
Meta-prompt first
Ask AI to write the prompt for you. Better than the one you would've typed.
Get local, get specific
City, season, audience, main road. Context = the difference between generic and postable.
Reputation is a revenue stream
Every unanswered negative review is a lead that went somewhere else. Draft in Claude first.
Agents are the buzzword
Last year AI answered questions. This year it completes multi-step work on its own.
Stay relevant — keep learning
Don't be the person still on dial-up. The tools are shifting class-to-class. Keep digging in.
remember this from year 1?
One hour saved per employee per day adds up fast. With Cowork in the picture, that math gets even friendlier.
the copy-paste prompts.
Tap the gold button on any prompt to copy it. Swap the [brackets] with your details and go.
local seasonal post
Specific, human, postable — no edits required.
the negative review response
Acknowledge · Empathize · Redirect. Never admit fault.
reusable job-posting prompt
Ask AI to write the prompt you'll use every time.
the difficult caregiver conversation
Captures your tone and the key details before you write.
new client onboarding sequence
Turn one good email into Day 1 / Day 3 / Day 7.
30-day content calendar
Turn what worked into a repeatable month.
how to talk to the robot.
The ACDQ rule — remember this?
act
Tell it who to be. "You are a world-class email-marketing specialist…"
context
The more context you give — audience, goals, constraints — the better the output.
deep think
Literally say "think deeply about this." The quality jumps noticeably.
question
Ask it to ask you questions before answering. You'll get far better results.
four LLMs, four vibes.
Different tools for different jobs. The operators we lean on — and what each actually does best.
claude
Best for: Long documents, strategy, deep analysis, nuanced writing.
Superpower: Reads 200K tokens (that's a whole book).
chatGPT
Best for: Images, voice, quick research, coding, video.
Superpower: Creates images, browses the web, talks back to you.
gemini
Best for: Email summaries, Drive search, YouTube breakdowns, trip planning.
Superpower: Lives inside Google — already knows your calendar, inbox, docs.
copilot
Best for: Outlook drafts, PowerPoint, Excel formulas, meeting notes.
Superpower: Works inside the Microsoft apps you already use every day.
claude at a glance — the models
the three claude products — pick the right one
translation for franchisees
ChatGPT = your operator. Best for speed, versatility, execution.
Claude = your analyst. Best for deep thinking, long documents, structured reasoning.
meet Cowork — Claude on your desktop.
Every AI tool has worked the same way: type, copy, paste, repeat. Cowork breaks that loop. Tell it what you need. Step away. Come back to finished work on your real files.
turn one-offs into templates.
Every time you get a good output, ask for the reusable version. That's how you compound. Here are real Home Helpers franchisee moves.
the 5-star workflow.
Families choose care the way they choose dinner: they read first. Every unanswered negative review is a lead that went somewhere else. This is a workflow, not a trick.
three things. this week.
Not someday. Not when things calm down. This week. Tap each box as you go.
things you might be wondering.
what's the difference between an LLM and an agent?
An LLM (large language model) answers your questions — Claude, ChatGPT, Gemini, Copilot. An agent takes it further: it plans, acts, and completes multi-step tasks on its own. Last year AI answered. This year AI does.
what should my AI policy actually say?
Permission slip, not detention slip. Cover: what data stays in-house, what's okay to put into AI tools, how to verify outputs, and who to ask when in doubt. Encourage experimentation within guardrails.
is AI video really usable right now?
Yes. Google's Veo generates video from a text prompt — huge implications for recruiting and marketing content. Start playing with short clips.
what's chatGPT's new glow-up I should be using?
Memory (it remembers your tone and processes across threads — game changer), Custom GPTs (one per role: onboarding, ops, marketing), image uploads, voice + vision, live Python & Excel for real-time analysis.
lindy — what is it?
An AI-driven outbound + revive engine that finds, enriches, and emails franchise candidates with human-level context. Great for re-activating stalled CRM leads and booking qualified calls. (Also: Lindy's been answering Ingrid's email.)
fun thing to try tonight?
HubX AI Video — in the App Store / Play Store. Turn photos into video. Pure fun, surprisingly useful for social content.
AI won't replace you.
someone using AI well will.
The question isn't whether you use it. It's how strategically.

