Click any question to read more.
Something we didn't cover? Write to us.
No. What we ask of you is to listen and reflect. You will not be asked to believe anything, recite anything, or affiliate with anything. The texts we draw from are ancient wisdom literature from the Indian subcontinent.
Darpan is an AI, and we want you to know that plainly. The difference is in what it is allowed to do. ChatGPT will happily play the guru: it invents verses, invents citations, forgets you between chats, and agrees with whatever you say. Darpan cannot quote what is not in its corpus. Every verse it hands you is a real one, retrieved from the texts and cited by chapter and verse so you can check it. It keeps structured, editable notes on what you are working through, so each session builds on the last. And it is tuned to reflect rather than flatter: it talks with you first, and saves its one composed line for the end.
No. Adhyatm is a companion, not a clinician, and it is not a substitute for professional care. Nothing it says is diagnosis or treatment. If you are in crisis or thinking about harming yourself, please reach a qualified human. In India: Tele-MANAS at 14416 (24/7, free), iCall at 9152987821, or the Vandrevala Foundation at +91 9999 666 555. The app also points to helplines when its safety layer picks up signs of acute crisis, but do not wait for a prompt; reach out whenever you need to.
Not a word. Devanagari is shown for those who can read it; transliteration and an English reading accompany every verse. Most users only ever read the English.
As long as you need. Darpan, the talk-first mirror, stays with you for the length of one sitting. A session can be three turns or thirty. When you tap Close, the mirror reads everything you sat with and hands you a single reflection plus a verse.
Meditation apps teach a technique: breathing and noticing. Adhyatm has two parts. Darpan is a place to speak into when something is on your mind, and be met by the steadiest version of yourself. Vyom is a place to release what isn't for the mirror, anonymously and ephemerally. Neither is better at the breath. They are doing a different thing.
Faint starter prompts sit beneath the composer: "Anxious about something," "A relationship is heavy," "I can't sleep," "A choice I have to make." Tap one to begin from. They are starters, not constraints.
Vyom is the anonymous side of Adhyatm. Some things you need to put down without making them a conversation. Vyom is where they go. Write a few sentences; they live for 24 hours, may be read by strangers in that window, and then disappear. You read one stranger's words in return. No comments, no likes, no feed. The relief is in being read, not in being answered.
Yes. No display name, no avatar, no profile. Releases are not tied to your account in any user-visible way: no one who reads what you wrote can find you, write back, follow you, or block you. Our database does store the link internally, so we can delete your releases if you ask, and so the report flow can remove a release that breaks the covenant. The link is not exposed and is not shared.
Two layers. First, a classifier reads every release before it posts. Words that point at self-harm route to a crisis flow with helplines and an offer to sit with Darpan instead. Words that contain contact information (phone numbers, emails, social handles) are blocked. Second, every release card has a quiet report button. After three reports, the release auto-hides while a human reviews it. We err on the side of removal.
Yes. Vyom is one door; you do not have to walk through it. The Home screen offers entries to both Darpan and Vyom; you can use Adhyatm only for Darpan and never see a release. Vyom does not push notifications. Nothing you do in Darpan ever crosses over into Vyom.
The Bhagavad Gītā in full. The thirteen principal Upaniṣads. A broad curated selection of the Mahābhārata, including the Vidura Nīti. The Rāmāyaṇa's Ayodhyā and Araṇya kāṇḍas. The Yoga Sūtras. The Aṣṭāvakra Gītā. The Brahma Sūtras. The Chanakya Nīti. Curated selections of the Atharvaveda, Manusmṛti, and Yājñavalkya Smṛti. About thirty thousand verses in total, and the library will grow.
It depends on the text, and we tell you which is which. For the Bhagavad Gītā, you are reading published translations, named when shown. For the other texts, the English reading is composed by Adhyatm, drawn from the literal sense of the Sanskrit and shaped to the sentence you brought; when our reading is wrong, we will rewrite it. The Devanagari is always the canonical text from public-domain editions, and the chapter and verse number is always given so you can find other translations and the commentaries.
Software trained on the texts themselves reads what you sat with and finds the line that most resembles the question you asked. Then the same software composes the reading and the reflection. We curate the corpus by hand; the per-session match is automatic. This is the honest description of what is happening, and we'd rather you know.
Because the match isn't always right. When a verse lands far from where you actually were, you can ask the mirror to try again, or carry the one you got. Sometimes the line that doesn't fit yet is the one fitting later. We read the cases where the verse felt far and use them to keep tuning the retrieval.
No. We never train a model on what you say, and we do not pass session content into any training pipeline, ours or anyone else's. The language models that run inside Adhyatm are commercial services (primarily DeepSeek for composition and Google Vertex AI for classification + transcription), used under paid API terms that exclude training on your content.
It is sent to our servers, transcribed via Google Vertex AI, and discarded. We do not retain the audio. The transcript stays in your account (it is what the mirror reads) and you can delete it at any time.
Yes. Write to privacy@adhyatm.co and all of it (sessions, transcripts, Notes, Vyom releases, the account itself) is permanently deleted within seven days. An in-app delete control and a bulk export to JSON ship with the open beta; if you need either sooner, the same address works.
In the privacy policy: what we keep, what disappears, who else touches your data, and how deletion works. It is short, and written in plain English.
Yes. The whole thing is free during our beta. Sessions, Vyom, the verse library: no card, no trial timer, no upsell ladder.
Probably not. The model APIs, the database, and the work of editing all cost money, and Adhyatm has to pay for itself eventually. We have not decided the shape of paid yet, and we don't want to. The right decision will be made with the people who use Adhyatm, when there are some.
There will be a path. Adhyatm is for anyone who wants to sit with the texts; the door does not get smaller because someone's wallet is. Write to us when the time comes.
Android first, then iOS. The Android beta opens gradually on Play Store, as we let new practitioners in. iOS arrives once Android is steady. There is no web app yet; Adhyatm is built for a phone in a quiet moment.
Sign-in will be by phone number: we send a one-time code by SMS, you enter it, and that is enough to begin. No password, no email required. This flow rolls out as the beta opens; the earliest testers get their accounts set up by us directly.
Because composing the reading is real work. The model is reading the transcript, retrieving a verse from the library, and writing a sentence that fits both. A short wait is the honest cost of a thought-out reply. A close should take well under a minute; if it drags past that, tell us, and we will tune it down.
Right now there is nothing to cancel. When there is, you will cancel in two taps and revert to free. Your Notes, your sessions, and the verses you carried will all remain unless you ask us to remove them.
Free during beta. There is no card to enter to begin.