01
Execution governance and human authority
Mandate, delegation, clearance, refusal and the boundary where intent becomes consequence.
Independent research on intelligent action
Researching the conditions under which intelligent systems may act.
VALO Research studies authority, evidence, execution and institutional design for intelligent systems that create consequences in the real world.
Our central questionHow can intelligent systems become more capable without weakening legitimate human authority, accountability or the ability to explain what happened?
Research areas
Five connected fields.
The work begins before product design. We study the institutional, technical and evidentiary conditions required for legitimate action.
02
Evidence, provenance and verifiability
What a system observed, what remains unknown and how claims can be independently checked.
03
Agent institutions and system design
Roles, responsibility boundaries, memory, escalation and durable organizational structure for agents.
04
Local, edge and physical intelligence
Governance when agents, sensors and actuators operate locally, offline or near physical consequence.
05
Societal resilience and long-term governance
How institutions preserve responsibility, knowledge and adaptive capacity as intelligence scales.
Active programs
Research designed to be tested.
Each program is organized around explicit questions, artifacts, experiments and limits rather than broad claims.
Authority at the Execution Boundary
Formal and operational models for determining whether a proposed action is legitimate at the moment before it becomes real.
ActiveContinuous Evidence and Integrity
Observation, provenance, immutable receipts and post-execution evidence loops that distinguish what is known from what is inferred.
ActiveGoverned Agent Institutions
How agents can be organized into bounded roles with independent evaluation, authorization, enforcement and verification.
ActiveEdge Intelligence and Physical Action
Local authorization, offline receipts and governance for systems that act through machines, devices and industrial environments.
DevelopingPublications
A versioned research record.
Reports, working papers, technical notes, experiments and position papers are published with status, version, date, authorship and limitations.
Public work should remain readable, citable and open to correction.
The publication library separates released research from product material and makes the maturity of each document explicit.
Open the publication library →What VALO Is
A current architecture overview. It is a living technical note, not a peer-reviewed paper or certification claim.
Read the note →Publication status is explicit: working note, preprint, technical report, peer-reviewed work or archived version.
Open problems
Questions worth working on together.
These are research invitations, not claims that the problems are solved.
Q1
What constitutes sufficient evidence before consequential action?
We need testable standards for completeness, freshness, provenance and uncertainty across changing operational contexts.
Q2
How can authority remain legitimate under autonomous delegation?
Delegation must scale without silently expanding mandate, losing accountability or turning technical capability into authority.
Q3
How should systems explain failure across organizational boundaries?
External providers, incomplete observability and physical systems make causality difficult. Evidence must show the first observable break without overstating certainty.
Q4
What governance belongs locally at the edge?
Low latency, limited connectivity and physical consequence require local control without isolating the system from institutional oversight.
Independent Research Collaborations
Independent architectures. Defined boundaries.
VALO Research can work with external architectures without absorbing their ownership, identity or commercial freedom.
Aurora-Lens × VALO Research
Margaret Stokes
Independent Architect & Creator of Aurora-Lens
Aurora-Lens is an independently owned architecture for governed evidence admission, persistent PEF state, provenance, authority and freshness, contradiction and unresolved-reference preservation, text and voice communication admissibility, and lawful continuation.
VALO contributes authenticated identity, mandate and delegation, execution context, final authorisation through REHT, enforcement and verifiable receipts.
Asymmetric boundary: Aurora NON_ADMIT cannot be reversed. Aurora ADMIT does not grant execution authority, and REHT may still refuse.
StatusAgreed non-binding architectural framework.
PhasePhase 0 documentation prepared.
ObligationNo current operational delivery duty for Margaret.
Commercial objectiveAcquisition remains Aurora-Lens’s primary objective; no acquisition by VALO is implied.
Collaboration
Independent scrutiny improves the work.
We invite researchers, reviewers, universities and technical practitioners who can challenge assumptions, test claims or contribute methods and evidence.
Research collaborator
Affiliated researcher
Independent reviewer
University partner
Benchmark contributor
Visiting fellow
Research to practice
The research is broader than any one product.
VALO Research develops and publishes the underlying questions, models and evidence. Operational products are kept on separate product surfaces so research maturity and commercial claims remain distinguishable.
REHT
REHT translates execution-governance research into an operational decision layer between intent and execution.
Go to REHT ↗