Intent‑Validated Non‑Verbal Computing
Manipulus is an open‑source research project exploring a future beyond the Command Era of technology. Instead of issuing explicit commands (apps, buttons, wake‑words), Manipulus investigates how human presence, gaze, and gesture can be translated into validated intent — locally, privately, and invisibly.
This repository documents the concept, architecture, and prototype direction of Manipulus. It is not a consumer product, and it does not collect or transmit personal data.
To liberate humans from the Command Era of technology.
We believe computing should adapt to humans — not the other way around. A home, device, or system should respond naturally to where you are, what you’re attending to, and what you intend — without forcing you to reach for an app, speak a command, or surrender privacy.
Manipulus aims to become an Intent‑Validation Operating Layer:
- Invisible by default
- Local‑first and private
- Activated only by sustained, verifiable human intent
Manipulus is an open‑source Non‑Verbal Communication (NVC) platform.
You can think of it as:
External Telepathy for the 99%
It explores BCI‑like capabilities (attention, intent, context) without implants, surgery, cloud dependence, or biometric storage.
At its core, Manipulus answers one question:
“Is this movement random — or is this intentional?”
Only when intent is validated does the system act.
- ❌ Not a gesture‑control toy
- ❌ Not always‑listening
- ❌ Not cloud‑dependent
- ❌ Not a surveillance system
- ❌ Not a consumer‑ready product
Manipulus deliberately avoids twitchy, reactive automation. No action occurs without temporal consistency and contextual confirmation.
Traditional automation reacts to events.
Manipulus reacts to validated intent.
Intent is inferred only when multiple signals align over time, such as:
- Gaze direction
- Skeletal posture
- Gesture persistence
- Spatial context
- Duration thresholds
Example (conceptual):
If gaze is maintained toward a defined zone and a gesture is sustained beyond a validation window → intent is confirmed → action is allowed.
This reduces false positives and preserves human agency.
Manipulus progresses through small, verifiable prototypes whose sole purpose is to test whether intent can be distinguished from noise using non-verbal signals — locally and privately.
These prototypes are intentionally minimal, disposable, and exploratory. Details are abstracted to avoid premature coupling to specific hardware, vendors, or implementations.
Manipulus is designed around strict constraints:
- No biometric databases
- No facial recognition
- No identity profiling
- No cloud streaming by default
- No data resale
Raw video is processed ephemerally for inference and is not stored.
This repository contains no keys, credentials, endpoints, or production secrets.
This repo may include:
- Concept documentation
- Architectural diagrams
- Prototype logic descriptions
- Research notes
- Experimental code (clearly labeled)
Anything that could compromise privacy or security is intentionally excluded.
Manipulus is early‑stage research.
- APIs may change
- Concepts may evolve
- Nothing here should be considered production‑ready
The purpose is exploration, validation, and open discussion.
Manipulus is built in public to invite critique, not hype.
If you’re interested in:
- Non‑verbal interfaces
- Local‑first AI
- Human‑computer interaction
- Post‑app computing
You’re welcome to observe, question, or contribute thoughtfully.
Manipulus is a research and experimental project.
It does not make medical, cognitive, or behavioral claims. It is not a brain‑computer interface. It is not intended for surveillance, monitoring, or behavioral enforcement.
Any demonstrations are conceptual prototypes only.
Manipulus explores what computing could feel like when technology learns to wait for human intent — instead of interrupting it.