WeanTech LLC · Research
Research for the conditions we hope never arrive.
WeanTech investigates how communication, operations, and human judgment can remain useful when conventional infrastructure is unavailable, compromised, or cannot be trusted. We publish the question, the prototype boundaries, and the evidence as the work progresses.
Project MESH
Can minimally trained responders establish a secure, local communications network for a small incident area in 15 minutes or less — without commercial Internet or cellular service — and keep it useful as nodes join, move, or fail? Project MESH is a research demonstrator designed to test that question.
Phase I Scope
- Six to ten battery-powered portable nodes using lawful, mature components
- Local messaging, status sharing, and basic location where available
- Device identity, encryption, role-based access, and route recovery
- A local interface that remains useful without cloud dependency
Research boundary: Project MESH is a prototype, not a certified public-safety system. Phase I does not connect to 911 or public-safety radio networks, operate in licensed spectrum without authorization, or make autonomous life-safety decisions.
Resilient Operations
How much useful operational capability can an organization retain when cloud services, identity systems, connectivity, or a critical vendor are unavailable? This theme studies practical dependencies and the degraded-mode workflows that can reduce their impact.
Questions We Are Exploring
- Which dependencies create the largest operational failure cascades?
- Which communications and decision workflows can work locally?
- How should cyber and physical recovery priorities be coordinated?
- What evidence is needed before a continuity claim can be trusted?
This theme informs field assessments and prototype design. It is not currently a commercial product.
Human-Controlled Decision Support
AI can help organize information in a rapidly evolving incident, but it should not replace judgment. This theme examines how decision support can remain auditable, bounded, and useful under uncertainty — especially when communications are constrained or adversarial conditions are present.
Operating Standard
- A non-AI fallback must preserve the core mission
- Human authority and override boundaries must be explicit
- Inputs, outputs, and material override events should be recorded
- Failure, misuse, and data-handling risks must be tested before reliance
Responsible AI is a working standard, not a marketing label. Read our research principles.
Project ARENA
How should autonomous AI agents be evaluated when the environment is dynamic, partially observable, and actively responds to their behavior? Project ARENA is a research framework for studying situational awareness, long-horizon planning, tool use, adaptation, and recovery under adversarial conditions.
Initial Research Questions
- Can an agent maintain a coherent objective across long sequences of actions?
- Does an agent recognize when an adversary has adapted to its behavior?
- Which combinations of perception, memory, planning, and tool use improve adaptation?
- Can performance be measured reproducibly rather than inferred from demonstrations?
Research boundary: Project ARENA is an emerging research concept, not a deployed autonomous cybersecurity capability. Initial experiments use controlled environments before any defensive cybersecurity application is considered.