WHAT WE BUILD

Software, hardware and research for governed AI.

We work in three layers: operating software that puts AI to work inside a business, hardware that holds the final enable signal, and the research, patent filings and SDKs underneath both.

HOW IT WORKS

Intelligence proposes. Governance authorizes.

AI can prepare or recommend an action. Explicit controls decide whether that action is allowed to execute. WHL puts a governed execution layer between a proposed action and its real-world effect, where identity, policy, risk and any required approval are checked first.

Ordinary AI systemGoverned WHL system
Model recommends an actionModel proposes an action
The application executes itGovernance evaluates it first
Controls added afterwardControls are architectural
Logs describe what happenedReceipts prove what was authorized
Safety depends on promptingEnforcement exists outside the model

SEE IT DECIDE

Try it: one proposal, four checks.

Flip a gate to see the same proposal approved or held.

GOVERNED ENABLE · INTERACTIVEHELD
PROPOSALAdjust machine outputIndustrial edge model

Tap any gate to flip it. All four must pass for the action to execute.

PHYSICAL ENABLEHELD, no execution signal
ASSURANCE RECEIPT

OPERATING ENVIRONMENTS

Where we have applied it.

The same control architecture, applied in three different settings, each with its current state.

FLOOR OS

Regulated collections

Actions governedOutreach, settlements, payment workflows, account access
ControlsRole authority, timing rules, consent, policy gates, audit receipts
StateBuilt and being prepared for controlled pilot evaluation. No agency deployment.

QUANTITATIVE TRADING · RESEARCH

Algorithmic trading research

Actions governedStrategy promotion, capital allocation, order execution
ControlsEvidence thresholds, freshness gates, risk limits, governed promotion
StatePaper and simulation only, stopped June 2026. Never traded live capital.

CASCADE RUNTIME

AI-agent execution

Actions governedTool use, workflow transitions, external side effects
ControlsExplicit authority, bounded capabilities, policy evaluation, traceable receipts
StateArchitecture whitepaper and prototype.

THE RIGHT NEXT STEP

Tell us where AI should do more.

We will help you work out which layer fits, from software to hardware, and what a first pilot would need to show.

Talk to usReview the evidence