Aura: five AI rooms, not one assistant
A single general assistant answers a career decision, a bad night’s sleep and a physics chapter in exactly the same voice. Aura splits that into five named characters, each with its own room, its own colour and its own job — and answers in Hindi, Hinglish or English.
The problem
One assistant, one tone, every question.
- A general assistant flattens everything. The same even, agreeable voice answers a hard decision and a homework question.
- Agreeableness is not advice. Ask for a recommendation and you often get a balanced summary of every option — which is not a position.
- Context bleeds. One long thread mixes your mood, your studies and your work until none of it is useful.
The five rooms
Pick who you need today. Switching rooms changes the voice, the colour and what the app is good at — and nothing carries over unless you want it to.
- The Council — five sharp, unsentimental minds sit with your problem, and one answers at a time. You get a position, not a committee-flavoured average.
- The Mentor — choose who is sitting across from you and the advice arrives in their voice, ego and conviction included. Useful precisely because it is not neutral.
- The Psychologist — slower and warmer, more interested in the question under your question. A one-tap mood check-in builds a picture over weeks.
- The Teacher — send a chapter photo or a PDF, get it explained properly with diagrams where a diagram is the honest answer, then ask to be quizzed and answer with a tap.
- The Poets — four of them, answering in verse.
How it is built
Aura is a React + Tailwind app wrapped for Android with Capacitor, so the web build and the app build come from one codebase. The character work is prompt and interface architecture rather than five separate models: each room carries its own system behaviour, its own constraints on tone and length, and its own visual identity, so the difference is felt before a single word is read.
The design decision that matters most is what does not happen. Rooms do not share context by default, so your mood log never leaks into your study session. And the mood log is written to the device, not to a server — the most sensitive thing in the app is the one thing that never leaves the phone.
Where it runs
Aura opens in the browser at aura.lzworth.in, signs in with Google and is free to use, in Hindi, Hinglish or English. There is an Android build as well — by the app’s own admission it runs a build or two behind the web version and still has rough edges, which is stated on its home page rather than hidden.
FAQ
What makes Aura different from a normal AI chatbot?
It is five characters, not one. Each has its own room, its own voice and its own job — a council that gives you a position, a mentor with an ego, a psychologist who listens, a teacher who quizzes you back, and poets who answer in verse. Rooms do not share context unless you want them to.
Does it work in Hindi?
Yes — Hindi, Hinglish and English.
Where is my mood data stored?
On your device. The mood check-in log is written to the phone, not to a server.
Is there an Android app?
Yes, alongside the web app. Aura’s own home page notes that the Android build runs slightly behind the web version.
Who built Aura?
Vansh Kashyap — full design and build, web and Android. The site credits it directly: “Designed and built by Vansh Kashyap.”
Want an AI product that is actually designed, not just wired up?
Most AI apps are one text box over an API. Tell me what yours should feel like.