MAANAS Intelligence Kernel
Right now, in this exact second...
someone is screaming silently.
AI-generated narration with automatic scrolling. Our research vision, with clinical benefits and privacy protections still to be validated.
SILENCE
But the crisis lives in the silence.
Our project asks whether optional technology could complement self-reported wellbeing and help people seek support between visits. It has not established that passive signals reliably reveal distress.
MANAS
Can technology help us notice what goes unspoken?
Investigate
Study patterns that may relate to wellbeing.
Explain
Explore understandable, transparent outputs.
Protect
Evaluate local processing and data safeguards.
Nudge
Research optional, culturally informed prompts.
Sustain
Evaluate whether ongoing support is useful.
Privacy as a
Research
Priority.
The MAANAS Intelligence Kernel explores local processing and federated learning as ways to reduce data exposure. These are design goals, not a guarantee of anonymity or device-only processing. Privacy protections must be verified for each implementation.
Read Our Privacy ApproachThe MAANAS Intelligence Kernel.
Built for Everyone to Build On.
We are building the MAANAS Intelligence Kernel: the research platform at the heart of our work. The web application, iOS app, and watchOS apps are our initial beachheads—applications built on the kernel to explore its capabilities and learn from specific use cases. Interactive demos illustrate these experiences; clinical benefits remain unvalidated.
Model Context Protocol
Research integration through @maanas/mcp-server explores access to experimental signal summaries. These outputs are not validated distress-risk estimates or clinical assessments.
Real-Time Streaming APIs
WebSocket and REST interfaces explore streaming facial, pulse-related, and acoustic features alongside experimental WLOP fusion outputs. Clinical calibration, accuracy, and suitability for care remain to be evaluated.
Client SDKs & Reference Apps
Swift, TypeScript, and Python SDKs and reference applications support technical exploration. Deployment with real users requires separate privacy, safety, and regulatory review.
Build with the MAANAS Intelligence Kernel
Explore our full technical architecture, mathematical explainers, and API specifications in the official whitepaper.
Our research goal: earlier access to support.
We have not established that MANAS prevents crises. Future studies must measure potential benefits, missed concerns, false alarms, and unintended harms.