The first time I tried journaling, I downloaded a popular American app that asked me, very earnestly, to "set an intention for the day." I sat there at 8am in Koramangala, having slept three hours, with a manager review at 11, and I closed the app.
It wasn't that I didn't want to journal. It was that the app was speaking a different emotional language. It assumed I had time. It assumed my problems were tidy. It assumed if I just "manifested" hard enough, the joint family WhatsApp group would stop asking me when I was getting married.
The gap
Indian mental health is having a moment. A widely cited Lancet study on India's disease burden estimated that one in seven of us lives with a mental disorder. Therapy waitlists in the metros have stretched to months. The number of mental-health apps in the App Store has doubled since 2022. And yet, almost every app I tried, from the meditation ones to the AI chatbots, felt like it had been imported. The prompts assumed nuclear families. The mood graphs assumed weekends were for relaxing, not for showing up to four cousins' functions. The crisis screens listed US suicide hotlines.
There's nothing wrong with importing tools. There's a lot wrong with importing tools that don't translate. Mental health is not a language-agnostic problem: the things that wind us up, the things we won't say out loud, the very way we describe an emotion (we say "tension" for what an American might call "anxiety") all live inside our culture.
The job of a journal isn't to fix you. It's to give you a room of your own in which you don't have to perform.
What we wanted
We wanted three things, and we wanted them stubbornly:
-
Privacy that was real, not a marketing claim. "Your data is safe with us" is the most overused phrase in tech. We didn't want a privacy policy that said this. We wanted an architecture where the words you write are encrypted on your device before they ever reach us: what we store of your writing itself is ciphertext, and no court order can turn that back into your journal. That promise covers your writing, not everything Mann holds: a short plaintext preview of each entry, your mood, and a few other things the app needs to work sit outside it, and the privacy page spells out exactly what those are. The key is generated on your phone and held in the device keychain. We never receive it, so we have no way to read your entries or move them for you. Your writing is readable only in flight, never stored: once for a safety check that runs on every entry so Mann can recognise a crisis, and again if you're within your daily reflection allowance or you tap Go Deeper (Premium only). We'd rather say that plainly than let you assume more than we deliver.
-
An AI that understood Indian life. Not as a feature flag. Not as a "regional language pack." Built into the prompts. When Mann reflects on your entry about your mother's expectations, it knows what izzat is. When you write about the EMI on your first car, it doesn't ask you about your "discretionary spending." When you write in Hinglish, switching mid-sentence the way we actually talk, it follows.
-
A journal, not a chatbot. A chatbot is a session. You ask, it answers, the conversation evaporates. A journal is a story you're building over time. Three weeks from now, when you write something that echoes what you wrote in October, Mann should remember and call back. That is the entire point.
What we got wrong (and fixed)
Our first prototype was beautiful and useless. We built a perfect onboarding sequence that asked twelve questions before letting you write anything. We tested it with five friends. Four of them never made a single entry.
Turns out, the friction we'd added in the name of personalization was killing the very behaviour we wanted. So we cut onboarding to four screens. Then three. Then we made the third one optional. Now, fifteen seconds after you open the app for the first time, you're writing.
Our second mistake was over-reflecting. The first AI version responded to every entry with three paragraphs of analysis. People wrote less because they felt watched. We dialled it back: now Mann writes one measured reflection when you finish an entry, not three paragraphs of unsolicited analysis. It listens first.
What's still hard
Three things keep us up:
- Tone. The line between "warm" and "saccharine" is thin, and AI has a habit of crossing it. Every prompt change gets tested against the same twelve scripted scenarios before it ships, never against your entries, which stay encrypted and unreadable to us.
- Crisis detection. Sometimes a journal entry is a cry for help. We've trained Mann to surface iCall and Vandrevala numbers when it sees concerning language, but we know we'll get this wrong, sometimes both ways.
- Pricing. ₹199 a month is less than two cappuccinos in Bangalore but more than half a day's pay in Patna. That's exactly why journaling itself stays free and unlimited, whether or not you ever pay us anything.
What's next
We're publishing more of these: short pieces about journaling, about how AI fits into mental health (and where it shouldn't), and about the specific shape of stress in modern Indian life. Subscribe in the app or follow along on the blog. And if you're holding more than you can show, the journal is here. No judgment. No performance.
दिल की बात, सिर्फ़ मन के लिए।
Want to start journaling tonight?
Mann is free to download. Unlimited journaling, no credit card.
