AI Engineering: Advent of AI with goose Day 9

Day 9: Gift Tag Generator - AI Recipes, Automation, and a Fully Scalable Tag System

Today with AI engineering I was not only about beautiful design and UI. It was about building a complete, automated, parameter‑driven generation system powered by goose Recipes and Python. A pipeline that can produce multilingual, QR‑enabled, metadata‑rich gift tags on demand.

I was cooking with goose! Recipes and all...

This challenge pushed me deeper into AI- powered architecture, structured parameter mapping, and scalable content generation than any previous day, recipe architecture, and automation workflow.

The Challenge: Build a Fully Automated Gift Tag System goose, automation, the recipe system, the generation pipeline, and the engineering

The goal wasn’t just to output a tag.

It was to build a repeatable, extensible system that could:

• Accept structured parameters
• Map styles to logic
• Generate multilingual content
• Produce consistent outputs
• Embed dynamic QR codes
• Generate multiple tags in batch
• Output standalone, portable files

This is where goose Recipes shine. It did!

What is a Recipe: Recipes are gooses reusable workflows that package extensions, prompts, and settings together. Share proven workflows with your team and reproduce successful results consistently.

https://block.github.io/goose/docs/guides/recipes/

The Goose Recipe: The Brain of the Entire System

The heart of is the RECIPE-gift-tag-generator.yaml. It's a declarative AI workflow that defines:

Parameter-driven generation
The recipe accepts structured inputs like:

• recipient_name
• language
• tone
• gift_description
• style
• qr_url
• version_metadata

This turns the tag generator into a configurable engine, not a one-off script.

Style mapping logic
The recipe maps each tag style to:

• A gradient family
• A typography set
• A tone‑appropriate poem style
• QR border rules
• Layout metadata

This is where the “intelligence” lives — the recipe decides how the tag should look and feel based on parameters.

Dynamic content generation

The recipe handles:

• Multilingual greetings in English, Spanish, French, and Italian.
• Shakespearean-style micro‑poems
• Tone adaptation (formal, playful, heartfelt, humorous)
• Conditional QR embedding
• Version metadata injection

This is AI engineering at its best.

Standalone Output
The recipe outputs fully self-contained HTML files with:

• Inline CSS
• Inline JS
• Embedded metadata
• No external dependencies

Meaning: the tags work offline, print cleanly, and can be shared instantly.

Python Automation: Batch Generation at Scale

The second half of the system is the generate-all-tags.py script which is a lightweight automation layer that:

Loads structured tag configurations

Each tag is defined as a Python dictionary with:

• Style
• Language
• Tone
• Gift
• QR URL
• Recipient metadata

Calls the Goose recipe programmatically

The script feeds each configuration into the recipe, generating complete tags in one run.

Integrates real QR codes

Using the QR Server API, the script embeds scannable QR codes tied to each recipient’s message.

Produces consistent, reproducible outputs

Every run is deterministic with same inputs, same outputs. This is essential for production workflows.

This is the kind of automation that scales from 4 tags to 400.

The Gallery: A Showcase, Not the Core
The gallery (gift-tag-gallery.html) is simply a beautiful UI visual wrapper for the generated tags.

The real engineering win is that the gallery doesn’t need a framework, build system, or backend — it just displays whatever the recipe and Python script produce.

The gallery is proof that clean, structured generation unlocks effortless presentation.

Why Day 9 Matters (Technically)

It demonstrates the power of declarative AI workflows

Goose Recipes let you define what you want, not how to code it.

It shows how AI can be used for structured, repeatable content generation

Not just text but full documents with metadata, logic, and conditional rendering.

It blends AI with traditional scripting

• Python handles automation.
• Goose handles intelligence.
• Together, they form a scalable system.

It proves that small, well-designed recipes can replace entire microservices

This tag generator could easily be:

• A festival tool
• A corporate system
• A school event generator
• A marketing automation asset

Success Metrics

• All parameters supported
• Four styles generated
• Multilingual content working
• QR codes scannable
• Batch generation successful
• Recipe logic validated
• Metadata included
• Fully reproducible outputs

It was a systems AI engineering challenge wrapped beautifully in UI.

My Final Thoughts
This AI engineering challenge showed me how powerful goose Recipes become when paired with structured parameters and automation. With a single YAML file and a Python script, I built a scalable, multilingual, QR-enabled gift tag system that feels like a real internal tool.

Day 9: Complete. System: Automated. Holiday Magic: Delivered.

This post is part of my Advent of AI journey - AI Engineering: Advent of AI with goose, Day 9.

Follow along for more AI Adventures with Eri!