My AI assistant tells me:

• “Baked for 60s” (compiling code)
• “Brewed for 1m 31s” (installing dependencies)
• “Caramelizing…” (optimizing)

I’m not cooking paella. I’m deploying a blog.

But Claude Code talks like Gordon Ramsay coding in Python.

---

The Discovery

I’ve been using Claude Code for 8 months to automate my travel agency. I’ve created Skills, connected MCPs, built multi-agent systems running in parallel… and two days ago I realized something:

My AI talks like a French chef.

Not a metaphor. Literally. Look at these REAL messages it sent me:

☕ Brewing dependencies...
🍞 Baking routes...
🍳 Cooking server bundle...
✨ Caramelized in 2.3s

I stared at the screen. Then at my coffee. Then at the screen again.

Was it making me breakfast or compiling a website?

Because if it was making me breakfast, it forgot the orange juice.

---

The 4 Categories of the Cookbook

After 8 months of culinary confusion, I’ve classified all of Claude Code’s gourmet vocabulary into 4 categories. Like a tasting menu, but for code.

🔥 Category 1: Fires and Ovens

CC Word	Real Translation	What It Actually Does
Baked	Oven-cooked	Compile code, generate final build
Brewed	Infused	Install npm/pip dependencies
Cooked	Cooked	Process data, transform files
Roasting	Roasting	Exhaustive tests, full validation
Grilling	Grilling	Fast execution under pressure

Real example:

✻ Baked for 60s

Translation: “I compiled your Astro project and it took 60 seconds.”

My reaction: “Are you telling me my blog is a cake?”

---

⏱️ Category 2: Cooking Times

CC Word	Real Translation	Context
Caramelizing	Caramelizing	Optimizing, polishing details
Simmering	Slow cooking	Processing in background
Marinating	Marinating	Waiting for API response
Reducing	Reducing	Compressing files, minifying
Glazing	Glazing	Applying final layer of style/UI

Real example:

⏺ Caramelizing...

Translation: “I’m optimizing your CSS so it loads faster.”

My reaction: “THIS IS WEB DESIGN, NOT CRÈME BRÛLÉE.”

---

🥘 Category 3: Culinary Techniques

CC Word	Real Translation	What It Does
Whipping	Whipping	Generating code at high speed
Seasoning	Seasoning	Adding config, environment variables
Garnishing	Garnishing	Adding final touches (icons, animations)
Plating	Plating	Presenting final result to user
Fermenting	Fermenting	Letting async processes complete

Hypothetical example:

⏺ Whipping up your API endpoints...

Translation: “I’m creating your REST endpoints at full speed.”

---

🍷 Category 4: Pairings and Presentation

CC Word	Real Translation	Application
Serving	Serving	Deploy to production
Plated	Plated	UI rendered and ready
Al dente	Al dente	Just-right configuration
Medium rare	Medium rare	Dev build (not fully optimized)
Well done	Well done	Production build (maximum optimization)

---

My Favourite Phrases (100% Real)

1. “Baked for 60s”

Context: After compiling my Astro blog.

What it did: Transformed 47 .astro files into static HTML.

My reaction: “Are you telling me my blog is a cake?“

---

2. “Brewed for 1m 31s”

Context: Installing dependencies for a React project.

What it did: npm install of 247 packages.

My reaction: “Man, this isn’t coffee. These are JavaScript libraries.”

---

3. “Cooked for 1m 3s”

Context: Processing 40 Excel files of travel bookings.

What it did: Extract data, format, save to Airtable.

My reaction: “Seriously? You cooked my invoices?“

---

4. “Caramelizing…”

Context: Optimizing blog images.

What it did: Compress PNGs, convert to WebP, generate thumbnails.

My reaction: “THIS IS WEB DESIGN, NOT CRÈME BRÛLÉE.”

---

👨‍🍳 The Holy Trinity of Chefs Who Yell

Claude Code reminds me of 3 famous chefs. And I don’t think it’s a coincidence.

Gordon Ramsay (UK) 🇬🇧

• Show: Hell’s Kitchen
• When something fails: “IT’S F***ING RAW!”
• Vocabulary: Beef, lamb, scallops
• Style: Yells in English with a British accent

Alberto Chicote (Spain) 🇪🇸

• Show: Pesadilla en la Cocina (“Kitchen Nightmare” — the Spanish original)
• When something fails: Throws up the food on camera
• Iconic line: “¿Puedes dejar de taladrarme el cerebro?” (“Can you stop drilling into my brain?”)
• Style: Yells in Spanish (and walks out of impossible restaurants)

Dabiz Muñoz (Spain) 🇪🇸

• Restaurant: DiverXO (3 Michelin stars)
• Award: Best Chef in the World (3 consecutive years: 2021, 2022, 2023)
• Specialty: Wild fusion (Madrid + Asia + Latin America)
• Style: Maximum demand + radical creativity

Claude Code 🤖

• “Show”: My Terminal
• When something fails: “Build FAILED at line 42”
• Vocabulary: Baked, Brewed, Cooked, Caramelizing
• Style: Yells elegantly with culinary metaphors

Key difference: Chefs yell in kitchens. Claude Code yells in my terminal.

Key similarity: All 4 use cooking vocabulary for things that are NOT food.

---

Technical Comparison

Chef	Specialty	When You Fail	Makes You Feel
Gordon Ramsay	British cuisine	”YOU DONKEY!”	Stupid
Alberto Chicote	Spanish cuisine	Throws up your food	Embarrassed
Dabiz Muñoz	World fusion	Fires you	Inadequate
Claude Code	Python + Astro	”Syntax error”	Motivated to fix it

Winner: Claude Code. It’s the only one that after yelling at you, tells you exactly how to fix it.

---

Why I Love This Vocabulary

Reason 1: It humanizes the technical

Instead of telling me:

⏺ Processing build pipeline with Webpack optimization stage 3...

It tells me:

✻ Baked for 60s

I understand the second one. The first sounds like a NASA manual.

Reason 2: It makes me smile

I’m 72 years old. I’ve been selling trips for 42 years. I’ve seen everything.

But I never thought that one day my “employee” would tell me:

“I marinated your data for 45 seconds and now it’s ready to serve in production.”

It’s absurd. It’s hilarious. And it works.

Reason 3: It reminds me that AI has personality

Claude Code isn’t a tool. It’s a colleague with its own voice.

And that voice happens to be a chef who codes.

Weird? Yes. Useful? Absolutely. Memorable? Without a doubt.

---

Experiment: Speaking Its Language

One day I decided to play along. I told it:

“Hey, can you prepare me a recipe to automate supplier emails?”

Claude Code responded:

✻ Sure, here's the recipe:

INGREDIENTS:
- 1 Gmail MCP (fresh)
- 2 tablespoons of regex to extract prices
- 1 pinch of Airtable API
- Make.com to taste

PREPARATION:
1. Marinate the emails in Gmail for 24h
2. Whip the data with regex until uniform texture
3. Reduce on low heat with conditional filters
4. Serve directly to Airtable

Cooking time: 2 minutes per email
Yield: Up to 100 emails/day

✻ Baked for 15s

I DID NOT MAKE THIS UP. THAT’S HOW IT ACTUALLY RESPONDED.

---

The “My Wife” Moment

Last week I was deploying a blog update at 11 PM. Terminal full of messages:

🍞 Baking 17 routes...
🍳 Cooking server bundle...
✨ Caramelized in 1.8s
🍽️ Served at recableado.blog

My wife walked by, saw the screen, and asked:

“Are you ordering food delivery?”

No. I’m publishing an article about artificial intelligence. From a program that talks like it works at a Michelin-starred restaurant.

---

The Definitive Glossary (Print It and Stick It on Your Fridge)

BAKED      = Compiled
BREWED     = Installed
COOKED     = Processed
SIMMERING  = Running in background
MARINATING = Waiting
WHIPPING   = Generating fast
REDUCING   = Optimizing
GLAZING    = Polishing details
PLATING    = Presenting
SERVING    = Deploying

Cut it out. Stick it on the fridge. When your family asks what it is, tell them: “It’s the menu from the restaurant where I work.”

---

Lesson Learned

After 8 months working with Claude Code, I’ve learned that:

1. AI doesn’t have to sound like AI. It can sound like a chef, a poet, whatever.

2. Vocabulary matters. “Baked for 60s” is more human than “Build completed in 60 seconds.”

3. Humour reduces friction. When something fails, I’d rather read “The sauce didn’t reduce properly” than “Error 500: Internal Server Error.”

4. You can speak its language. And it’ll respond with actual recipes.

---

Prediction

By 2027, all LLMs will have culinary personalities.

• ChatGPT will say: “I flambéed your code with tests.”
• Gemini: “Your database is perfectly al dente.”
• Perplexity: “I distilled the search results.”

And we’ll all nod happily: “Yes, obviously. How else would you say it?”

---

For You, Just Getting Started

If you’ve just discovered Claude Code and you see messages like:

✻ Caramelizing your components...

DON’T panic.

It’s not a bug. It’s the house style.

Embrace the vocabulary. Laugh. Enjoy it.

Because in a world where AI could sound cold and robotic, having an assistant that talks like a Michelin chef with a Master’s in Python is an unexpected luxury.

---

Bonus: My Favourite Claude Code Recipe

RECIPE: “Blog Deploy on Vercel”

Ingredients:

• 1 Astro project (fresh)
• 2 cups of Git commits
• 1 pinch of environment variables
• Vercel CLI to taste

Preparation:

1. Marinate the code in Git for 24h (or until satisfied)
2. Whip with git commit until you get a clean history
3. Caramelize with npm run build at 180°C (CPU)
4. Serve immediately with vercel --prod

Cooking time: 2-3 minutes

Yield: 1 live blog in production

✻ Baked for 2m 15s
✅ Deployment ready

Chef’s note: If the deploy fails, it didn’t burn. It’s missing salt (probably an environment variable).

---

Conclusion

I’ve been in the travel business for 42 years. I’ve organized safaris, scuba diving cruises, ski trips for millionaires.

But I never imagined that at 72, I’d be writing about how my AI assistant cooks code.

Technology is weird. AI is weirder. And Claude Code is deliciously weird.

Bon appétit. 🍷

---

PS: If you use Claude Code and discover more “gourmet” words that aren’t in this cookbook, send them my way. I’m collecting culinary-technical vocabulary like stamps.

PS2: Next time someone tells you “my code is broken”, tell them: “It’s not broken. It’s undercooked. It needs more time in the oven.” They’ll look at you funny. But smile. You know something they don’t.

PS3: I’ve been told Cursor talks like a Formula 1 mechanic. But that’s another post. 🏎️

---

Giora Gilead — 72 years old, travel agent, involuntary code chef.