I recently built TaroMind AI, a web app for guided tarot readings:
At a surface level, it is an AI tarot product.
But what I actually wanted to build was something more specific:
a calmer, more deliberate reflective tool that uses tarot structure without feeling like a thin chat wrapper.
Most AI products in this space felt like one of two things:
- a generic chatbot with mystical wording
- a flashy tarot app optimized more for stimulation than reflection
I wanted something slower.
Something that felt closer to a ritual than a prompt box.
The product idea
TaroMind AI is built around a guided flow:
- set an intention
- choose an interpretive guide
- draw cards in sequence
- receive a reading
That pacing is intentional.
I did not want the product to feel like:
- type question
- get instant blob of text
- leave
The point was to make the interaction feel more grounded and a little more intimate.
The core product distinction
The part I care about most is the interpretive framing.
TaroMind is not just “AI generates tarot text.”
The product is built around two lenses:
- one leans more toward Western archetypes and Jungian-style interpretation
- one is more contemplative and influenced by Eastern thought
So the same question can be approached in more than one way.
That is the experiment behind the product.
Not prediction.
Not certainty.
More like structured reflection through symbolic language.
What is in the current version
Right now the product includes:
- 3 complimentary readings for new users
- a guided tarot flow from intention to result
- 22 public Major Arcana card meaning pages
- a private journal for saved readings
- premium plans through PayPal subscriptions
I also wanted the public site to do real work, not just act as a shell around the app.
So the card library is public and indexable:
/cards/cards/[slug]
That lets someone discover the product through card meanings first, then move into a live reading later.
Why I made it
I like products that create a certain pace.
A lot of software is optimized to reduce friction at all costs, which is usually the right instinct. But sometimes a little structure is part of the value.
With tarot, I felt that removing all pacing made the experience worse.
Tarot already has a built-in symbolic grammar:
- question
- spread
- card
- meaning
- interpretation
So instead of replacing that structure with a chatbot, I wanted to preserve it and build around it.
That became TaroMind.
The stack
The app is built with:
- Next.js 14 App Router
- Supabase for auth and persistence
- OpenAI for interpretation generation
- a local JSON-backed knowledge layer for card meaning context
- PayPal for subscriptions
One practical constraint I cared about was making the app feel like a product, not just a wrapper.
That influenced a lot of small decisions:
- public pages have real metadata and structure
- pricing is visible before sign-in
- private features stay behind auth
- the reading flow is designed, not just generated
The part that was surprisingly tricky
The hardest part was not generating tarot text.
That part is relatively easy.
The harder part was making the whole thing feel coherent:
- the landing page
- the ritual flow
- the card pages
- the upgrade path
- the tone of the interpretations
For a product like this, credibility does not come from saying “AI” more loudly.
It comes from whether the experience feels internally consistent.
What I am trying to learn
I am still figuring out what this product really is for people.
A tarot skeptic might see it as a reflective writing tool.
A tarot reader might see it as a guided interpretation companion.
Someone else might just see it as an interesting AI product experiment.
That is part of why I wanted to share it here.
I would love feedback on
- whether the flow feels clear or too slow
- whether the two-lens framing feels meaningful
- whether the product feels distinct from a generic AI wrapper
- whether the card meaning pages make sense as a public entry point
If you want to try it:
If there is interest, I can also write a follow-up post about the technical side:
- how I handled SEO for a new App Router site
- how I made the pricing page public without exposing payment actions
- how I structured the card library as a real content surface instead of a dead-end marketing page
Top comments (0)