Aatman

Aatman is an app that helps you transcribe your life. You record anything you want with voice notes and it stores the information in such a way that you can query for anything like people, events, actions items, feelings etc. Essentially it helps you get rid of thinking and writing every thing that is important to you. Which is the reason why most of us give up recording our lives.

The idea isn't to noke make another productivity tool something to help you but like a soundboard for your thoughts which regularly checks upon you and evolves with you.

Why?

The answer is simple. Modern life is run like a startup. Every aspect of life needs to be taken care of discreetly because well...our conventional support systems aren't working. We are on our own. But we are not equipped to handle all of it ourselves. Reason why there were support systems in the first place. Now we've got reddit and it's just overwhelming.

And honestly, after a good day of work, I just don't want more work that need the same level of decision making. So anything that asks for more than my immediate thoughts as input is already out of question. Voice is the least frictional way of useful input. So it should be enough for the system to be useful. This is the core mission of the app.

How it works?

Your record voice notes. The app generates the transcript. The transcript is anaylzed and a knowledge graph is extracted from it using LLMs. The knowledge graphs is then stored as vector embeddings. The knowledge graph contains entities, relationships and properties. Entities are things like people, places, things, etc.

Whenever the user issues a voice command, it is first categorized either as a query or a thought. If it's a query, a strcutured query is generated and a fairly complex search is performed, using semantic search, relationship traversal and building possible graph paths. The graph paths are then scored and ranked to collect the best possible entities and relationships. The final results are fed to the LLM to generate a natural language answer.

Desired Vision

Right now the goal is to evolve it into a generative co-intelligence system that grows along the user and prompts them back with questions, actions and thoughts. It should be able to reason across the different aspects of the user's life and help them make decisions.