Who may authorise
Can a person prove they are entitled to act without handing over more than the purpose needs?
In public source
Explore Identity & VEID →Research portfolio
One research question, four programs
Can people and increasingly autonomous AI systems share infrastructure and make consequential decisions within an architecture that keeps human authority verifiable and prevents either side from exercising unchecked control? Each program answers part of it. The work is grounded in published artifacts: the open VirtEngine repository, granted patent AU2024203136B2, the 103-page DSEMA specification, Bosun's source, and the signed constitution. Questions are stated as questions; we do not invent results.
Read the research synthesisTry one task, end to end →NIST AI RMF alignment
Connected research
Each part answers a different question.
The programs are designed to depend on one another. Each role below links to the program that carries it; the map shows the proposed dependencies, none of which is yet an integration.
What agents may do and change
Can agents adapt while every action and amendment stays inside verifiable limits?
Published design
Explore DSEMA →Where work runs and is paid for
Can resources be acquired from independent providers on terms anyone can inspect?
In public source
Explore VirtEngine →What supervision looks like in practice
Which lessons from supervised agents doing real engineering work transfer to the wider design?
In operating use
Explore Bosun →Interactive research mapChoose a participant
Intended beneficiaries
People
People choose goals, question results, and need meaningful access to services. The economic hypothesis asks whether participation can remain possible when human labour is less in demand.
- Contributes
- Objectives and demand → agents
- Depends on
- Useful services ← agents
Human–machine symbiosis: the research synthesis →Interactive learning library
01Programs
Four programs, one architecture, one lock
Each program carries its own technology, patents, and documentation. All four are parts of the same research architecture, and all four are held under the same public-benefit lock.
Program 01 — decentralized cloud
VirtEngine
A decentralized cloud computing marketplace connecting tenants who need compute with providers who lease capacity — identity, encryption, escrow, and settlement enforced by the chain itself.
- stack
- Go · CometBFT · Cosmos SDK
- license
- Apache-2.0, open source
- patent
- AU2024203136B2 — granted to 12 May 2044
- identity
- VEID — privacy-preserving
Program 02 — trustworthy AI
DSEMA
The Dynamically Self-Evolving Multi-Agent architecture: collectives of specialized AI agents that improve themselves under verifiable on-chain control and an immutable constitutional safety layer.
- subject
- self-evolving multi-agent AI
- status
- research — patent pending
- control plane
- blockchain orchestration
- safety
- constitutional safety layer
Program 03 — autonomous engineering
Bosun
A fully open-source, experimental control plane for autonomous software engineering: executor routing with failover, automated PR lifecycles, self-healing recovery — the applied track that informs the DSEMA research.
- status
- experimental — open source
- license
- Apache-2.0
- install
- npm install -g bosun
- home
- bosun.engineer
Program 04 — verifiable identity
Identity — VEID (Verifiable Electronic Identity)
Verifiable Electronic Identity: multi-factor verification encrypted on-device, scored by chain consensus, presented as zero-knowledge proofs of attributes — with identity.org.au as the citizen-facing service.
- layer
- x/veid · x/mfa · x/veid/zk
- patent
- AU2024203136B2 — granted to 12 May 2044
- service
- identity.org.au
- privacy
- on-device · zero-knowledge
02Research topics
Six questions the programs share
Cross-cutting research streams that run through the programs. Each topic page states the open question, the approach, current work, and the verifiable artifacts behind it.
01 · VirtEngine
Decentralized identity
Can a person prove who they are — strongly enough for finance-grade infrastructure — without surrendering their documents, biometrics, or behaviour to a central database?
Explore the topic
02 · DSEMA
Multi-agent AI safety
If a collective of AI agents can modify its own models and structure, what makes it safe, accountable, and stoppable?
Explore the topic
03 · VirtEngine + DSEMA
Confidential computing
How can tenants run sensitive workloads on hardware owned by strangers — with confidentiality and integrity they can verify rather than assume?
Explore the topic
04 · VirtEngine + DSEMA
Protocol governance design
How do you govern valuable shared infrastructure so that no amount of capital, cleverness, or patience can capture it?
Explore the topic
05 · VirtEngine + DSEMA
Incentive mechanism design
What economic mechanisms keep a decentralized marketplace honest — pricing, escrow, settlement, reputation — when every participant is anonymous and self-interested?
Explore the topic
06 · VirtEngine + DSEMA
Privacy-preserving verification
How much can you verify about a person, a machine, or a computation while learning nothing else?
Explore the topic
For the institutional context behind the research — why the foundation is obliged to publish it and how the outputs are locked to the public — see the mission and the constitution explorer.