A Different Kind of Technology

Our business model is simple: you pay us directly for a reading, and that sustains the work. Because we don’t rely on outside investments or advertising, we don’t need to monetize your attention.

There are no ad-tracking pixels here. No data brokers. No Google Analytics or Meta tracking. What we do measure to understand our own site, we host on our own servers. We don’t have a cookie-tracking banner because there is nothing to hand off. If you open your browser’s developer tools, you can verify: nothing here is reporting your actions to an advertiser.

What you say in a reading, and what comes back, are yours and yours alone. Unless you’ve explicitly asked us to review them, or there’s a safety or legal emergency, or clear abuse of the service, no one on our team reads your conversations. We treat access as a rare exception, not the norm—so you can be fully honest, and bring your real questions instead of your careful ones.

Notice we never asked for your last name? Why would we need it? For a reading, exact birth date, time, and place are actually necessary, but even those act as a kind of fingerprint—precise enough to identify you—so we keep it, and all your readings and other personal data, securely encrypted at rest. For everything else that a reading doesn’t need, and which we don’t sell to advertisers anyway, we don’t ask for it, and we don’t keep it even when a login or payment provider tries to hand it to us. If it’s not essential for what we do, we don’t want it. This is the surest way to keep the container safe.

We do give you the option to add much more personal info about yourself, but all of it is entirely optional, and again all encrypted and only leaves our servers when essential for the services we provide.

There is only one time your data leaves our servers: to generate a reading. Even then it goes only to a small set of AI providers — and we only work with ones we can hold to agreements that bar them from training on what you share and require them to delete it within a short time (we seek the shortest retention we can find). We send only what a reading actually requires, and only for as long as it’s needed. The rest stays with us. That’s the one door out.

We looked at the alternatives: hosting the models ourselves, to close this door entirely. So far, the quality of wisdom and reliability possible that way simply isn’t there. In the meantime, the honest tradeoff we’ve made here is the best one we’re aware of.

For a simple reading you can dump your basic chart into an AI and discuss (what most AI-horoscope apps online do now). But an AI is a language model—it is famously bad at calculation. If left on its own, it will happily invent complex transits or confidently put your Moon in the wrong sign (in addition to inventing things).

To address this, we built our own astronomical engine, using the same high-precision ephemeris data NASA’s Jet Propulsion Laboratory uses to navigate spacecraft. The placements are calculated, not guessed. Crucially, whenever the AI needs a complex number—like the exact angle of an aspect, or the full arc of a transit over several years—it doesn’t guess. It asks our engine to work it out. The reading itself is a constructed process, using multiple AIs that re-pull and verify the data against each other. This leads to results that have much greater precision, but can also go much wider and deeper than the data you began with. (More on how readings are built: How Readings Work.)

Most AI is built to agree with you. An off-the-shelf model has absorbed the entire internet—which means it holds both profound human insight and every shallow, fatalistic pop cliché ever written. If you just ask for a reading, it defaults to a noisy, flattening average of all of it. Tell a standard chatbot what you dread, and it will happily pluck a random planet out of that noise and hand your anxiety right back to you as destiny. Most days this means mediocre results, but now and then it can be outright dangerous.

To get to the wisdom, you have to build better maps. Through countless hours of testing, we have explored this space firsthand. From both our failures and moments of blinding insight, we learned to navigate these waters, and made maps of them. The system we’ve built is a guided journey based on those maps—one that steers away from the shallow noise, around the dark maelstroms, and toward the deep places where the mystics, the elders, and the myth-makers stored their most profound and useful wisdom.

On top of this, the AI is grounded in a carefully curated well of foundational archetypes. We require it to operate from a deliberate value system, marrying the best of ancient astrological knowledge with a clear-eyed understanding of the human mind.

Because it is rooted in this architecture, the reflections our readings offer are entirely different. They are deeply compassionate, but they will also gently push back on the thought traps we all fall into rather than validate them. Pointed at depth rather than engagement, we built this to tell you not just what you want to hear, but what you might actually need. (More on that: Sacred Silicon.)

Because we have spent considerable time building with these tools, we know exactly where their edges belong. A reading is not therapy, and it is never a substitute for professional mental health support. If something you’re carrying is heavier than a reading should hold, our system is designed to say so plainly and point you toward real human help. (If you are ever in crisis, please reach out to a human—in the US, call or text 988; globally, findahelpline.com lists free, confidential support near you.)

But most days aren’t a crisis. When you are simply navigating life’s chapters, looking for meaning, untangling a complex dynamic, or seeking a quiet space to reflect, we are here. For all the days when you just need a wise, steady mirror, this is exactly what we built Starwisdom for. (See our full Commitments.)