I used an AI tarot app for 'life advice' for a week — here's what actually happened

Last updated: June 2026 · By Thomas Løvaslokøy, founder & builder of BrainLoop. AI-assisted in drafting and fact-checked against the sources listed at the end. Tarot here is for entertainment and reflection — not prediction, diagnosis, or decision-making.

I'm going to be completely honest with you from the first sentence, because I built the thing I'm about to review: BrainLoop's Tarot is two simple parts wearing a mystical coat. Part one is a shuffle — the code takes the major-arcana cards, randomises them, and deals you three. Part two is an AI: once the cards are down, Claude writes you an interpretation of each one plus an overall reading. So it is "AI tarot" — but knowing exactly what that means matters, because neither the random draw nor the language model can see one second into your future. When I opened it every morning for a week and asked it for "life advice," I wasn't testing prophecy. I was testing something more interesting: whether an AI-written reading of three random cards can still be useful.

Here's what actually happened.

What the app actually does

You press "Draw Three Cards." Three cards turn over — Past, Present, Future — drawn at random. Then the AI does its part: it interprets each card in its position and ties them into a short overall reading, in fluent, personal-sounding prose. Thirty seconds, no prophecy, no robe required. The important thing to hold onto: the AI is handed three randomly-chosen cards and nothing else about you — it doesn't know your life, it's writing a plausible reading from a coin-flip.

On day one I drew The Tower in the "Present" slot. The old version of this app would have shown me a fixed one-liner; the AI instead wrote a paragraph about foundations I'd built on assumptions that might need re-examining — specific-feeling, well-written, a little uncanny. I felt that jolt of oh, that's relevant. And the builder part of my brain said: relevant to what? You have a hundred things going on. Of course a paragraph about "shaky foundations" lands.

That jolt has a name — and the AI makes it stronger.

Why an AI reading feels like it knows you

In 1949, psychologist Bertram Forer gave his students what they believed was a personalised personality analysis. In reality, everyone got the exact same paragraph of vague, flattering statements. Asked to rate how well it described them, they averaged 4.3 out of 5.¹ We are spectacularly good at taking a generic statement and feeling it was written for us — the Forer effect (or Barnum effect), and a big part of why horoscopes and tarot feel uncannily accurate.¹²

Here's the twist worth understanding in 2026: a large language model is, in a sense, a Forer-effect engine. It's extraordinarily good at producing fluent, plausible, broadly-applicable text that sounds tailored to you. So an AI tarot reading doesn't escape the Barnum effect — it supercharges it. The fixed one-liner was easy to see through; a warm, articulate AI paragraph is much harder. Add confirmation bias — you keep the hits, forget the misses — and by day three I was nodding along to readings I knew had been generated from three random cards. Knowing the trick didn't switch it off. That's how strong it is.

The honest part: as prediction, it's useless

Let me say the quiet bit loudly. An AI interpreting a random draw cannot tell you what will happen, or what to do about your money, your health, your relationship, or any decision with real consequences. The model isn't perceiving your future — it's generating plausible language from three cards a shuffle handed it. If a reading ever seems to tell you to quit your job or end a friendship, the AI isn't talking — you are, and you've found an articulate excuse to avoid owning the choice. For anything that actually matters, talk to a real person who's qualified to help. AI tarot is theatre with a very good script — not a therapist, not an oracle.

If anything, the more convincing the writing, the more firmly you have to keep the stakes at zero. So that should be the end of it — except the week genuinely surprised me.

What it was genuinely good for

The value isn't in the answer the reading gives. It's in the question it makes you sit with.

When The Tower came up, the useful part wasn't the AI's confident prose — it was that it nudged me toward: "Is there anything in my life right now built on a shaky foundation I've been ignoring?" A genuinely good thing to ask yourself on a Tuesday, and one I'd never have reached on my own. A reflection prompt I choose for myself tends to point where I'm already comfortable looking; a random card, dressed up by an AI into a vivid little narrative, points somewhere I'm not — which is exactly where the useful thinking hides. Used as a daily "draw, read, then journal the question it raises" ritual, the week was quietly worthwhile. Not because the cards or the model knew anything, but because they made me stop and think for ninety seconds before the day swallowed me.

How to use an AI tarot app without fooling yourself

• Read it as a question, not a verdict. "What might 'upheaval' be pointing at?" beats "upheaval is coming."
• Remember a fluent reading is still just plausible text from three random cards. Polish is not insight.
• Draw once and sit with it. Redrawing until you get a "good" reading is fortune-shopping, not reflecting.
• Pair it with journaling. Write the question it raised and your honest answer — that's where the value actually lives.
• Keep the stakes at zero. The better the writing, the stricter this rule. The moment it steers a real decision, stop.

The verdict after seven days

As a window into the future: nil — and to its credit, BrainLoop's Tarot never claims otherwise. As a thirty-second morning ritual where a shuffle and a sharp little AI hand you a poetic prompt to reflect on, then get out of your way? Genuinely pleasant, and a few times it knocked loose a thought I needed. I went in knowing every card was a coin-flip and every reading was generated on the spot, and I still got something out of it — which, honestly, is the most accurate thing tarot has ever told me about human nature.

Draw your three cards here — just remember that you, not the AI, are the one giving them meaning. That's not a bug. That's the whole point.

---

Sources

Footnotes

1. Barnum effect (Forer effect). Reference overview of Bertram Forer's 1949 "fallacy of personal validation" study, in which students rated identical generic personality descriptions as personally accurate at a mean of 4.3/5; the effect helps explain belief in horoscopes and tarot — and applies equally to fluent AI-generated text. https://en.wikipedia.org/wiki/Barnum_effect ↩ ↩²

2. The Decision Lab. Why do we believe vague, general statements describe us accurately? — The Barnum Effect. Explainer on how the Forer/Barnum effect and confirmation bias make generic statements (horoscopes, tarot, personality quizzes) feel individually tailored. https://thedecisionlab.com/biases/barnum-effect ↩