In prototyping
A personal life-OS research prototype - documents, contracts, lab results, nutrition, wearables, media, and manual input in one grounded assistant. Not a commercial product; shared as concept work.
Concept
Brain is a personal research project: a life-OS built around CORE and a connected data graph. CORE combines an LLM, skills, agents, cloud runtime, and storage - not a single chat window. Instead of answers trapped in chat history, the stack works over your sources: files, contracts, lab results, nutrition logs, media libraries, manual input, global statistics and reports, and (later) wearable signals. The point is to explore grounded recall and compounding context - not to sell another chatbot.
Capabilities
A personal second-brain research project - documents, contracts, labs, nutrition, wearables, media, and manual input in one grounded life data graph.
01
LLM core connected to personal sources, not chat history alone
02
Exploring wearable sync for sleep, workouts, and heart-rate context
03
Manual input, media, and labs sit in one graph with docs and contracts
Product surface
Each block maps to a planned data source or surface in the product graph.
01Feature
Core inputs are documents, contracts and agreements you need to re-find, medical lab results as structured context, nutrition tracking for daily grounding, media libraries, global statistics and reports as structured summaries, and manual input you log on the fly. Wearables join as a live stream once device sync lands - sleep, workouts, and heart rate sitting next to the same graph.
02Feature
Manual input captures what you did and what was on your mind - short logs, decisions, and reflections that would otherwise vanish into chat history or nowhere. Those entries become searchable source nodes, so Brain can answer from both durable files and the day-to-day trail of intent.
03Feature
Media libraries become another source node - events, places, people, and visual context the assistant can search and cite alongside text. Planned multimodal ingest keeps media linked into the same graph as documents, so questions can span a receipt photo, a contract PDF, and a lab PDF without hunting folders.
04Feature
Fitness watch integration is planned so Brain can read daily sleep, exercise, and pulse data without manual export. That stream becomes another source node - useful for questions that need both life context and physiological trends - and stays optional until the sync path is ready.
05Feature
Lab results and food tracking feed the same assistant as documents and contracts. The goal is one place to ask grounded questions across paper trails and body signals - without scattering context across apps. Nothing here claims clinical advice; it is personal context for the operator's own second brain.
System design
Eight data sources feed CORE - LLM, skills, agents, cloud, and storage as one stack. Grounded recall and answers flow back to you.
[External System]
Documents
Files and exports as immutable sources
[External System]
Contracts
Agreements and decisions on record
[External System]
Media
Photo, video, and personal media libraries (planned)
[External System]
Manual input
Logged actions, decisions, and reflections
[External System]
Statistics
Aggregated stats and periodic reports
[External System]
Lab results
Medical analyses as structured context
[External System]
Nutrition
Food tracking and daily intake
[External System]
Wearables
Sleep, workouts, heart rate (planned)
Brain
[Software System]
[Component]
LLM
Inference
[Component]
Skills
Tools
[Component]
Agents
Workflows
[Component]
Cloud
Runtime
[Component]
Storage
Graph
[Person]
You
Operator
Interest
If Brain is useful as a research direction, leave your email - we may share notes when the work advances. Not a waitlist for a product sale.
