Overview: Kalidos – Style Quiz with AI And Seasonal Lookbook
Some links in this article are affiliate links to products I actually use. If you sign up through them, I may earn a small commission — at no extra cost to you.
The inspiration for a style quiz with AI came from my partner. What initially seemed like a small, proof of concept quickly became the start of something else. “Kalidos”, from the Greek kalos (beauty) plus eidos (form, appearance) – 2 of the same roots that combine to form “kaleidoscope“, a view of beautiful forms – tests the premise that personal style is better described as a combination of independent traits, than being assigned into a handful of pre-defined buckets.
The core positioning: shape trends, don’t follow them
Whether this holds up as a product mechanic – not just a philosophy – is what this experiment is designed to find out.
What Kalidos Actually Is
Kalidos is an AI-assisted style exploration concept – it maps an aesthetic identity across 16 independent facets: silhouette, colour palette, occasion, values, footwear, fragrance, and more. Rather than assigning labels like “Minimalist” or “Bohemian”, it treats those facets as coordinates – an AI model interprets the specific combination with a personalised narrative; another generates a shareable illustrated card, and – since the quiz is designed to recombine, not sort – it adapts to change.
In this first phase, there are two main components. First, the quiz itself – a free tier, plus a paid unlock, to generate your full “Style Kaleidoscope”. Second, a seasonal lookbook engine that generates and showcases the results for a cast of fictional personas, to validate the concept even when no one is actively taking the quiz.
The tech stack is layered on WordPress, with Formidable Forms for the quiz UI, n8n orchestrating Claude, Mistral, OpenAI, and Gemini (with visitor-selectable models and rendering styles on the paid tier), Google’s image APIs, Stripe for payment, and MailPoet for the mailing list.
Privacy was a design constraint from the start, not an afterthought. The system uses a pseudonymous identity model with unbundled consent, versioned profile history, and an automated retention schedule that actually deletes expired records. Decoupling style profiles from personally identifiable information means a partial database breach exposes a lot less – a practical consideration, not just a compliance checkbox.
I’ve written about parts of the underlying infrastructure in more detail elsewhere: automated WordPress publishing pipeline with n8n and Claude covers the publishing foundation; and simplifying n8n LLM routing with a provider abstraction layer explains how I manage multi-model routing without the workflows becoming unmaintainable.
The Style Kaleidoscope in Practice
The free experience requires no email or registration – the entry barrier is deliberately low. Answer 9 questions and you get a personalised narrative and image card, illustrated in Classic Croquis. The optional €8.88 paid unlock extends this considerably to explore different perspectives of your personal aesthetic:
| Free Tier | Optional Paid Unlock (€8.88) | |
|---|---|---|
| Style Facets | 9 core facets | Full 16-facet profile |
| AI Narrative Writer | Randomly assigned AI model | Choice of Claude, Mistral, GPT, or Gemini |
| Visual Output | 1 custom illustrated card | 1 custom illustrated card + moodboard |
| Illustration Styles | Classic Croquis (default) | Choice of 8 distinct display styles |
| Regenerations | 1 per week | Up to 3 per week |
That regeneration cap matters operationally, since every run has real API costs. Remove those limits and a single user could exhaust the budget in an afternoon. Rate-limiting and the optional unlock are needed to keep the experiment financially viable (the affiliate layer is technically functional; whether it will generate meaningful passive income is not yet proven).
The seasonal lookbook engine adds an ongoing dimension. It runs monthly, with occasional ad-hoc runs tied to real-world events, generating new looks for a small cast of fixed fictional personas – Ingrid, Malik, Sage, and Rio. Each run references the persona’s established style history, so their look visibly shifts over time – which is the whole point. A Style Kaleidoscope is meant to change, not stay fixed.
You can see this in action in the Kalidos Seasonal Lookbook.
Why Build A Style Quiz With AI
This started as a fun experiment, and style is a good litmus test – visual, low-stakes, and subjective enough that variation in the AI’s output can be interpreted as creative range rather than inconsistency. One thing Kalidos has certainly demonstrated so far is that letting 16 independent axes interact freely inside an LLM prompt produces genuinely surprising results.
However, the real output is the foundation for a reusable, domain-agnostic, modular pipeline: configurable trait axes feeding AI interpretation and image generation, with a free/paid tier split, a seasonal content engine, and a built-in affiliate layer. Swap the facets for career skills, interior preferences, or culinary tastes and the same architecture could generate personalised career roadmaps, room designs, or recipe collections.
Curious what your own Style Kaleidoscope looks like? Try the free quiz yourself: triple8labs.com/kalidos/quiz
More details can be found at the Kalidos experiment page.
