3 candidate company names — checked, not just guessed
A favorite, with a tagline
Your logo direction locked + a ready-to-generate image prompt
A working mental model of what Claude actually is
First, what we're working with.
Claude is an AI model — one of the best in the world. Made by Anthropic. Trained on most of what humans have written down.
You talk to it. It talks back.
You didn't come empty-handed
You already started.
You showed up with a name — No Worries Roofing — and a logo. That already puts you ahead of everyone who never gets past "someday."
So we don't start from a blank page tonight. We put what you've got to work — and find out if it's the strongest version of the idea.
Where we're heading · the whole arc
Chatbot
Tonight starts here. Sessions 1–4. Claude as a writer and analyst. You ask, it answers. Drafts, summaries, brainstorming. Where 80% of the value lives for most people.
Agent
Sessions 5 + 7. Claude with hands. It reads your inbox, schedules your calendar, builds you a working tool. Still you-in-control, but Claude can do things now.
Agentic workflows
After this course. Whole jobs running end-to-end — lead capture → reply → schedule → follow-up, all automated. Internal tools you'd never have hired a developer to build.
Each stage builds on the one before it. Skip ahead and nothing works. Master the basics and the rest gets easy.
Your turn · right now · paste this in Claude
I'm starting a roofing AND solar company. I'm
considering the name "No Worries Roofing."
Play devil's advocate. Score it 1-10 on each:
1. Does it say what we do? (roofing AND solar)
2. Right feeling for a big, scary, expensive job?
3. How crowded is it — other "No Worries" home
companies out there?
4. Easy to rank on Google + grab a clean .com?
5. Room to grow if we add services later?
Be blunt. Then tell me the single biggest weakness.
Read it together
What did it just tell you?
The good
"No worries" nails the feeling — peace of mind on a high-stakes purchase. That instinct is right. Keep it.
The catch
It drops solar — your whole bundle play. And "No Worries [trade]" is everywhere. Hard to rank, hard to own.
Now widen it · paste this next
Now give me 12 alternatives that keep the same
"peace of mind / stress-free" feeling but fix the
weaknesses you just found. Three groups:
- Tweaks that keep "No Worries" but add solar
- Same feeling, more distinctive
- Names that bridge "roof" and "sun" so the name
itself signals the bundle
One line each on why it works.
When a batch is mid
Don't start over. Refine.
Bad approach
Close the chat. Open a new one. Re-paste. Hope for better.
Good approach
Reply in the same chat: "These [3] are close. 5 more like them. Drop anything that sounds like a tech startup." Each round, it learns your taste.
Now let's look under the hood
What is this thing, actually?
You just generated a pile of names in five minutes. Time out — let me show you what's actually happening in there. It matters for what we do next.
Common misconception
It's not a search engine.
When you ask Google something, it goes find pages that match. When you ask Claude something, it doesn't go anywhere. There's no "looking up."
It's a prediction machine. It learned patterns. From those patterns, it predicts what a good answer looks like.
Try this
Fill in the blank.
What does that mean for you?
✓ Great at
Writing drafts. Editing. Summarizing what you give it. Generating options. Brainstorming. Anything where it's working WITH information you provide.
✗ Bad at
Recalling specific facts. Names. Dates. Statistics. Today's news. Anything where it has to "look something up" — it can't.
Important
It will lie to you.
Sometimes Claude makes things up. Confidently. Cites a fake book. Invents a statistic. Quotes someone who never said it.
"Same input → exact same output. Has to be right."
The total on an invoice. Square footage × price/sqft. Whether John is on Tuesday's schedule. Payroll. Tax. Sales tax in your county.
Code's home turf.
When AI "lies" about facts, it's because you asked a probabilistic tool to give you a deterministic answer. Wrong tool for the job — not a Claude failure.
Before we move on · privacy
Where does your stuff go?
When you chat with Claude, your message travels to Anthropic's servers
On Pro/Max consumer plans, Anthropic does not train future models on your chats by default
They do retain your conversations for ~30 days for safety review
Staff can see flagged chats — for abuse review, security, bug debugging
Of the big AI labs, Anthropic has the most privacy-forward defaults. We'll do a brand-by-brand comparison in Session 3 when you start uploading real documents.
Last concept before we build
Context.
Each chat with Claude is like a meeting. Whatever's been said in this meeting, Claude remembers.
New chat = new meeting. Claude has amnesia between chats.
Watch context build up
Three things to remember tonight
1. Predictor, not searcher
It writes from patterns it learned. It's not looking anything up.
2. Verify the facts
It can lie confidently. Double-check names, dates, statistics.
3. Stay in one chat
Context builds up. Each message makes the next better.
Quick check.
In your own words — what's the difference between Claude and Google?
Back to your shortlist · the big lesson
AI brainstorms. You verify.
Remember — it predicts, it doesn't look anything up. When it says ".com is probably free" or "that's not trademarked," it's guessing. Names and vibes: trust it. Hard facts: check them yourself.
Search the .com at any registrar
Google the exact name + your city
Your state's business-entity registry
Instagram / Facebook handle free?
Quick trademark gut-check (USPTO search)
Pick three.
Three that survived the checks — including "No Worries" if it's still standing. Write them on paper.
Next artifact
Now the tagline.
Pick your favorite name. We build a tagline for it together. The other names get taglines as homework.
Paste this in Claude
The company name is [NAME]. It's a roofing and
solar installer in Azle.
Brand vibe: regional, blue-collar pride,
family-owned feel. Not tech-y, not corporate.
Customers should feel like they're hiring someone
whose dad also did this work.
Give me 10 tagline candidates. Mix it up:
- Some short (3-5 words)
- Some longer (a full sentence)
- Some functional ("we do X")
- Some emotional ("how you feel after hiring us")
Tell me which lane each one is in.
Then: pick 2 you like → "combine the best of these into 3 more" → pick the one you'd put on a truck.
The big artifact tonight
Now — your logo.
Here's the move: we don't make Claude draw the logo — that's not its job. We use Claude to design the concept and write the exact recipe. Then you make the real picture in a tool built for pictures. Right tool for each job.
The concept · 30 seconds
What makes a logo work.
A wordmark is enough to launch — your name, the right font and color. The fancy icon can come later.
Readable tiny and huge — a phone icon and a truck door
One or two colors — no gradients stacked on gradients
Simple and ownable — not the stock clip-art every other roofer uses
Step 1 · the concept · paste this in Claude
I'm designing the logo for [NAME], a roofing +
solar company. Tagline: [TAGLINE]. Vibe: local,
trustworthy, blue-collar, family-owned — not
corporate, not tech-y.
Give me 3 distinct logo directions. For each:
- The concept in one line
- Wordmark style (the typeface feel)
- A simple icon idea that reads at phone-app size
- Color palette (2-3 colors, with hex codes)
- One thing to avoid
Keep it simple enough for a polo and a yard sign.
No clip-art.
Step 2 · the recipe · paste this next
Take direction #[X] and write ONE detailed image
prompt I can paste into an image generator to make
this logo as a PNG. Spell out the wordmark text,
typeface feel, layout, colors (hex), background,
and overall style. One tight paragraph, ready to
paste.
Save that prompt + the direction in your No Worries Roofing & Solar folder. Making the actual image is the next step — Ryan will walk you through the tool for that.
The real skill
Right model for the job.
Claude is your brain for words and ideas. For images, you reach for an image model — and even those don't behave the same. Half the skill is knowing which one to pick.
ChatGPT
Its image generator listens — feed it the recipe and you get close to what you actually asked for. Where we start.
Higgsfield
Bring the best version here to refine — variations, tweaks, polish until it's right.
Midjourney
Wildly creative and beautiful — but it does its own thing. Stunning images that often aren't what you asked for.
Behind this is a dial called temperature — low means it sticks to what you said, high means it gets wild and surprising. Different models, different defaults. Match the tool to the job.
Recap · the three rules
1. Predictor
Not a search engine. It writes from patterns it learned, not from looking things up.
2. Verify facts
It can lie confidently. Always check names, dates, statistics.
3. One chat
Iterate. Each message adds context. New chat = blank slate.