About the practice
Work you
can inspect.
A3GIS is an independent research and engineering practice. Its work moves between agent infrastructure, applied perception, assistive hardware and the methods used to test an idea.
A concrete starting point
A person.
A device.
A useful signal.
SmartCane makes the question physical. An ESP32 prototype senses distance, communicates alerts and supports remote monitoring and an SOS flow. A reading becomes meaningful through what a person can do with it.
Computer Vision approaches perception from another starting point: collaborative university work on face detection and estimation. The next ambition is task-focused models that fit the memory, latency and power limits of embedded hardware.
AEGIS-COGNITION asks where authority belongs when software acts. Proof-Driven Engineering brings the work back to evidence: define the requirement, inspect the result and retain the limits of what was checked.
The thread through the work
Seeing is only
the beginning.
A shared direction, built from independent projects.
Read the project stories →- 01
Sense the world.
Start with the information a device can obtain, and the conditions under which that information remains useful.
- 02
Fit the intelligence.
Choose what a model needs to recognise. Measure the hardware constraints before selecting an embedded target.
- 03
Give action a boundary.
Make responsibility, constraints and evidence explicit before extending an experiment into a system that acts.
How the practice works
Build the idea.
Question the result.
Public repositories make the software work inspectable. Project stories distinguish documented foundations, the owner-confirmed hardware prototype and proposed research. Collaborative work retains its credit.
The aim is to develop useful intelligence with a clear account of what exists today and what the next experiment should establish.
Start with a useful question →