VΛLIS, a new architecturefor autonomous computing.
Intelligence+Authority+Software+Physical systems
EXPLORE / NINE SUBSYSTEMS
VALIS DESKTOPILLUSTRATIVE DEMO · NO LIVE APPS
YOU
● ● ●DEMO
Choose the memory scope for this illustration
This decision affects the demonstration only.
NotesCalendarTasksApprovalsReceipts
DRAG TO ORBIT · SCROLL / PINCH TO ZOOM
The 3D architecture could not load. The written architecture below remains available.
VALIS / COMPUTING ARCHITECTURE
ILLUSTRATIVE CYCLE / NOT LIVE EXECUTION
INTENT → CONSEQUENCE → EVIDENCE
Trace one proposal through the machine. Select a stage to inspect it. No live approval or external action takes place.
THE AI RACE, REFRAMED
We do not need a race for supremacy. We need a machine worthy of our speed.
Artificial intelligence is accelerating—but raw intelligence is only power. Like the energy driving a Formula 1 car, it becomes useful when an integrated machine converts it into controlled movement.
VALIS / THE INTEGRATED MACHINECAPABILITY ≠ AUTHORITY
01 / SUPPLY
Intelligence supplies power.
Models, agents, humans, and tools prepare what could happen.
02 / SETTLE
The kernel converts it.
Authority, scope, and current conditions determine what may be released.
03 / ACT + OBSERVE
Consequence stays bounded.
The destination checks the release. Readback establishes what actually happened.
↶ EVIDENCE RETURNS TO THE NEXT PROPOSAL Recursion carries learning forward. It does not make authority unlimited.
EXPLORE THE ENGINEERING ANALOGY
Settlement kernel / Engine
The recursive settlement kernel evaluates the relationships required to release a proposed consequence.
VALIS is being built as that integrated machine for autonomous computation. At its center, the Recursive Settlement Kernel connects proposals to the conditions for authorized, bounded, and verifiable consequence.
The ambition is civilizational infrastructure for directing machine intelligence through a shared world—across organizations, networks, robots, institutions, and economies—without confusing capability with authority.
The defining question is no longer only, “Who can build the most powerful intelligence?”
Who can build the machine that allows intelligence to become consequence—without losing human authority over the world it creates?
An engineering analogy for the architecture, not a runtime execution trace. Formula 1 is used as an analogy; no affiliation is implied.
02 / COMPUTING, COMPOSED
Every era gave the machine a new power.
Compute. Remember. Connect. Learn. Act. VALIS brings these capabilities into an architecture that makes the conditions for action explicit. A model can prepare a deployment, a payment instruction, or a device command; the authority to carry it out remains a separate question.
1936ƒCOMPUTATION
The machine becomes formal.
Turing’s abstract machine gives computer science a general model of what can be computed.
Intelligence begins planning, calling tools, coordinating work, and reaching into physical and institutional systems.
NOWRSETTLEMENT
Ability meets legitimacy.
VALIS adds the missing layer: the conditions under which intelligence may become action.
03 / THE ARCHITECTURAL SHIFT
Intelligence becomes part of the machine.
A model can reason. A harness can sustain work. Tools can reach into the world. The settlement kernel connects proposed consequences to the authority and conditions that govern them.
The holographic machine above is a visual model of these relationships. Its animation is illustrative, not a live execution trace.
01MODELWhat can be inferred?02HARNESSHow does it keep working?03TOOLS · SKILLS · PLUGINSWhat can it reach?04MCP · A2AHow does it connect?05RUNTIMEWhere does work persist?RSETTLEMENT KERNELWho may authorize which exact consequence?06SHARED WORLDWhat actually changed?
04 / DIGITAL × PHYSICAL
One architecture. Many forms.
Software, agents, organizations, and devices retain their distinct roles. They connect through a shared pattern: prepare a proposal, establish its conditions, authorize the exact action, then return evidence.
This is the composability of the machine: more participants can join without capability alone expanding their authority. A service can prepare a work order, a person can authorize its scope, a device can execute within that scope, and the observed result can return to the shared work graph.
Digital and physical participants connect through the settlement kernel.
05 / FROM ARCHITECTURE TO WORK
What could you compose?
01 / SOFTWARE
A governed deployment.
An agent prepares a change. The responsible person reviews its exact target and scope—for example, one reviewed build for one staging environment. An adapter carries the released action to that destination; readback records the deployed revision. A production target or different build needs its own applicable authority.
02 / PHYSICAL SYSTEMS
A bounded device action.
A device request carries a specific target, operation, and time window—for example, a maintenance adjustment to one machine. A changed target or out-of-range setting cannot inherit the original release. Observed device state helps establish whether the requested adjustment actually happened.
03 / COORDINATED WORK
A workflow that resumes.
People and agents share dependencies, decisions, and evidence—for example, a supplier review followed by a purchase request and delivery handoff. Interrupted work resumes from recorded state. A completed task is not repeated blindly, and current authority is checked at the next consequence.
Illustrative application patterns. Deployment-specific integrations require their own verification.
Boundary demonstrator
Change the action. Watch authority stop.
A local illustration, not the VALIS kernel or a live approval surface. No purchase is made.
SUPPLIER / NORTH WORKSHOP · LIMIT / $200
Awaiting example approval
Bind a demonstration approval to the current order value.
06 / UNDERNEATH THE WORK
Different industries. The same consequence boundary.
VALIS is proposed as an underlying coordination and settlement layer wherever intelligence prepares actions across software, organizations, or physical systems. It is not a replacement for each industry’s applications, expertise, safety controls, or accountability.
FINANCE + INSURANCE
A reviewed financial operation.
An assistant prepares a payment instruction, claim adjustment, or account change. The accountable operator reviews the exact amount, destination, and scope. The system records the resulting transaction or refusal. The settlement layer does not replace financial controls or professional judgment.
HEALTHCARE + LIFE SCIENCES
Coordination with accountable decisions.
An agent prepares a scheduling change, record update, or laboratory workflow. Authorized staff review the specific action and data scope; destination evidence records what changed. Clinical decisions and patient-data permissions remain with the responsible people and systems.
MANUFACTURING + ROBOTICS
From a plan to a bounded device action.
A planning system proposes a production change or robot task. The release specifies the device, operation, limits, and validity window. The execution interface checks those conditions and returns observed state; physical safety systems remain necessary.
Explore more industries and markets
ENERGY + INFRASTRUCTURE
Control changes with a traceable basis.
An assistant proposes a maintenance window or equipment setting. The responsible operator reviews the target and allowed range. Device readback and incident records connect the decision to the observed result.
LOGISTICS + SUPPLY CHAINS
Coordination across organizational boundaries.
Agents prepare a shipment reroute, procurement order, or inventory transfer. Different parties retain their own authority. Exact destinations, quantities, and handoff evidence make the cross-system workflow inspectable.
SOFTWARE + CYBERSECURITY
A specific change to a specific environment.
An agent prepares a deployment, access grant, or incident-response step. A release binds the approved revision, environment, and permissions. Logs and destination readback distinguish an approved plan from an executed change.
PUBLIC SERVICES + INSTITUTIONS
Visible responsibility in shared systems.
Assistants prepare a service update, records operation, or resource-allocation proposal. The institution retains decision responsibility. Scope, affected systems, and evidence of the resulting action remain available for review.
EDUCATION + RESEARCH
Collaboration with permission and provenance.
Agents coordinate experiments, learning workflows, and shared research artifacts. Access, authorship, participant permissions, and release conditions remain explicit when a proposal becomes a published or operational change.
MEDIA + COMMERCE + CREATIVE WORK
Publication without losing the context.
An assistant prepares a campaign, product-listing change, or creative release. The relevant people review the exact content, audience, rights, and destination. Publication readback records the version that actually reached the public.
Illustrative application patterns, not customer deployments or sector certifications. Each integration needs its own scope, controls, and destination-level verification.
07 / THE HUMAN ORIGIN
Before infrastructure, a way of seeing.
VALIS emerged from Ricky’s fourteen-year inquiry into how meaning, identity, authority, and consequence arise through relation—across culture, language, science, institutions, and machines.
In 2012, finance and black-box trading made the question concrete: machine decisions could become consequential at speed while their reasoning and authority remained difficult for affected people to inspect.
Relation before infrastructure.
The inquiry widened through systems and cybernetics, community practice, science, culture, language, and art. As capable language models arrived, the relational framework gained a new experimental medium for mathematics, software, and repeatable tests.
The architecture follows that starting point: identities keep their distinction; authority remains situated; actions have conditions; and evidence connects a decision to what actually happened.
Ricky and Jesse’s conversations began as roommates in Newport Beach in 2012. Their technical dialogue resumed around the emerging architecture in September 2025, connecting the long inquiry to review, implementation, and evaluator design.
The architecture has an originator. Its development has a constellation: engineering, relational modeling, community, law, and responsible deployment contribute distinct perspectives.
RA
FOUNDER · INVENTOR · PRINCIPAL ARCHITECT
Ricky Aliae / Maymay
Founder and principal architect of VALIS. His work connects relational intelligence, systems research, and meaning-centered cultural practice, from relation as primitive to settlement calculus and institutional direction.
Engineering lead in AI product systems and Ricky’s oldest technical interlocutor. Jesse brings implementation discipline, adversarial review, evaluator design, and the bridge into production organizations.
Cofounder and Director of Intelligence Relations at Fostering Projects and Syzygy Laboratory. Alia contributed relational modeling, organizational coherence, and early testing.
Attorney, community convener, and co-author of Conspiracy of Dreamers. Renee connected the project to community, legal, technical, and institutional review.
An early strategic sounding board on responsible experimentation, public readiness, engineering operations, and the practical implications of the architecture.
Noetica / interpretive layer. Noetica helps turn admitted context and the work graph into shared understanding. It may interpret and propose; it does not become the source of authority.
Start with one real task
Where should an AI be allowed to act on your behalf?
Choose one task where software or an AI assistant would change something outside the conversation: release a software update, pay an invoice, change a customer account, or move a machine. That is the workflow we want to explore with you—from the first proposal to a result someone can verify.
You do not need a finished integration. Bring the way the task works today, who is responsible for it, and what would make you comfortable letting an agent take the next step.
01 / YOUR EXAMPLE
Bring the task and its stakes.
What should happen? Which systems would change? Who can authorize it? What must never happen? A short description or walkthrough is enough to start.
02 / THE WORKING SESSION
Find the decision boundary.
Together, trace where a suggestion becomes an action. Identify the exact change to approve, its limits, who can stop it, and the evidence needed to confirm the result.
03 / A CONCRETE NEXT STEP
Define what a useful test would prove.
The aim is a workflow map and scoped evaluation plan: what to test, what counts as success, which failures must be blocked, and what remains unresolved before a pilot.
For example / Releasing a software update
An agent prepares a release. Who lets it go live?
The agent proposes a specific version for a specific environment. A release owner reviews that change and its allowed window. If the version changes, permission expires, or the owner revokes it, the proposed release must be checked again before execution.
The evaluation compares the approved version with what reached the destination. A useful result includes both the successful path and evidence that an altered or expired request was stopped—not simply that the agent said “done.”
What you gain: a clearer division between what the agent may prepare, what a person must decide, what the execution system must enforce, and what evidence your team needs to trust the outcome.
This begins with a scoping conversation. The worksheet below helps you prepare; it does not run an evaluation or connect to your systems. Integration work and any pilot scope are agreed separately.
Review the published architecture and evidence. Distinguish design claims, simulations, reproducible implementation results, and destination-confirmed behavior.
We are looking to connect with teams bringing a concrete agent workflow, engineers building integrations, researchers who can challenge the architecture and its evidence, and partners interested in a carefully scoped pilot.
Tell us who you are, what you are building, which action needs a clearer authority boundary, and what you would want an evaluation to demonstrate.
Email info@recursivesettlement.com to begin a conversation. The evaluator and technical links open the existing public site. A demo result or setup page does not issue live-system authority; the worksheet above remains a private local download.