Research

Research directions.

How energy, connected devices and intelligent systems can remain understandable, useful and under human control.

A spatial landscape of connected configuration regions, with one reachable route highlighted in mint.

System behaviour

The shape of possibility.

Points are possible states; lines are permitted transitions.

What the laboratory studies

Questions shaping the work.

What should keep running when power is uncertain? How can a device explain its state? Who can stop an automated action? These questions connect the laboratory's research across physical and digital systems.

Energy continuity

Continuity when infrastructure is under stress

How can local energy storage and battery systems support continuity when centralized infrastructure is delayed, unstable, expensive, or unavailable?

Energy is becoming more local, electronic, and software-mediated. Homes, sites, devices, field operations, and intelligent systems increasingly depend on power decisions that ordinary users rarely see.

Why it matters

The question is not only battery capacity. It is what should continue operating, what should pause, what should recover first, and how people can understand the state of the system when infrastructure is under stress.

Future systems may need to behave less like distant utilities and more like recoverable local infrastructure.

Connected electronics

Physical objects that can explain their state

How should physical objects, rooms, cabinets, and devices carry state without becoming opaque, fragile, or uncontrolled?

As electronics enter cabinets, rooms, machines, vehicles, and everyday environments, physical state becomes part of software behavior. A device reports condition, accepts commands, grants access, records events, and sometimes coordinates with other systems.

Why it matters

A connected object should not become trusted merely because it is connected. It becomes useful when its state, authority, and recovery path can be understood.

Connected electronics may become part of trusted environments only when physical state and permission are designed together.

Cryptographic authority

Authority after action begins

When systems can act beyond a screen, where does authority live and how can it be recovered?

A sign-in proves that someone entered a surface. It does not always prove that a later action still belongs to the right intent, scope, or moment.

Why it matters

As software, devices, rooms, and agents prepare action, authority needs a stronger shape: signing, scope, duration, revocation, proof, and recovery.

Future environments may need personal authority devices, scoped delegation, recoverable keys, and audit trails that make permission understandable.

Intelligent systems

Help without hiding judgment

How can intelligent systems help without moving judgment out of view?

At first, intelligence feels like acceleration: a document is summarized, an operation is routed, a next step is proposed. Then a quieter question appears. If the system can continue on its own, where can a person inspect evidence, narrow the task, interrupt the path, or understand why the recommendation exists?

Why it matters

AI is entering documents, enterprise systems, public surfaces, and operational workflows at the same time. Without visible control, helpful automation can become hidden dependence.

The useful intelligent system may be the one that preserves judgment while reducing unnecessary effort.

Agentic frameworks

Bounded delegation

Why do agent frameworks need authority, context, tool, evidence, and operator planes instead of orchestration alone?

An agent framework can look capable when it routes tasks and calls tools. The harder question is what the system is allowed to know, what it is allowed to touch, which evidence it must preserve, and where an operator can intervene.

Why it matters

Agentic systems are moving from chat responses into tool use, multi-step work, generated surfaces, and shared operational context.

Future agentic systems may be judged less by how much they can do and more by how clearly they can explain what they were allowed to do.

Knowledge preservation

Understanding that future work can inherit

How does work become knowledge that future work can inherit?

Work leaves traces: notes, drafts, decisions, errors, proofs, questions, and follow-up. A team can keep all of them and still fail to learn.

Why it matters

A memory becomes useful when it is selected, checked, placed, and connected to later action. This matters more as AI systems produce more summaries and artifacts than people can comfortably inspect.

Future knowledge systems may need to carry evidence, decision history, review state, and next action together.

Operational visibility

State that becomes useful context

How can state from the physical world become readable without becoming careless surveillance?

Telemetry, traceability, and operational software matter because work often becomes unclear while it is still happening.

Why it matters

A machine changes condition. A material moves. A vehicle enters a route. A document waits for approval. A process step is skipped or delayed.

Future operations may need shared situational awareness that is purposeful, bounded, and recoverable.

Identity

Portable authority

What does identity mean after systems are allowed to act?

Identity is more than sign-in. As intelligent systems act across environments, research shifts toward custody, signing, revocation, and permissioned action.

Why it matters

Recognition is becoming easier, while permission remains difficult to explain across devices, agents, and organizations.

Identity may need to become a portable authority layer that can prove intent without overexposing the person.

Human consequence

The habits a product teaches

What habits does a product teach when it becomes part of daily life?

Every product teaches behavior. It changes what feels normal, what feels slow, what feels safe, what people trust, and what they stop checking.

Why it matters

Many technologies are intimate enough to shape default behavior before a person notices the new habit.

Future products should be judged not only by what they enable, but by what kind of human behavior they quietly train.

How inquiry becomes useful

Observe, frame, prototype, review.

A direction becomes stronger when the question stays visible, the evidence is not overstated, and the work remains small enough to learn from before it becomes a public claim.

  • ObserveLook for small shifts already happening in work, tools, teams, and decisions.
  • FrameName the tension before proposing the system.
  • PrototypeTest the behavior at small scale, with careful boundaries around data and access.
  • ReviewKeep the question open long enough to learn from what the prototype changes.

Further reading

Follow a question further.

Read about authority in systems that prepare action, consider a possible workshop interaction, or explore earlier software and connected-system work.

Follow the signal essaySee the permissioned roomExplore selected work