Construct the distinction
The input provides primitive channel equations and an uncertainty class. The compiler builds weighted measurement programs; it receives no catalogue of finished instruments.
MONTREAL.AI / SUCCESSOR Ω
Autonomous Enterprises · Activate. Evolve. Scale.
Build the observation.
Unlock the next decision.
The Discoverability Compiler constructs a measurement program that can answer a bounded question—and explains when the permitted observations cannot.
Experimental · executes locally in your browser01 / Construct and use
Your first instrument.
Choose a mission, construct its experiment, then acquire new synthetic readings and inspect the decision.
Construct a probe program that distinguishes six digital-module conditions despite shared offset and drift.
No wallet, credentials, model API or server execution. Cost units describe simulated acquisition work.
Loading the source-bound workbench…
Ready for a new execution
A capability you can inspect.
The compiler will synthesize legal contrasts, test every pair of incompatible decisions and select a minimum-cost program within the permitted grammar. Ambiguity and exhausted work stay visible.
—constructed experiments
—acquisition units / decision
—world decisions checked
| Experiment | Channel weights | Noise bound | Cost |
|---|
Review and settlement
Constructed programs remain REVIEW_REQUIRED. Independent operator admission, Fleet execution and settlement are not connected to this browser experiment.
Decisions, measurements and limits
No execution yet.
02 / Make the result falsifiable
What changed—and what did not.
Check the original question
A fit is insufficient. Every admitted world consistent with the complete observation must require the same decision. Touching uncertainty intervals remain ambiguous.
Earn inheritance again
A changed decision or nuisance model invalidates the earlier certificate. Reusable construction code survives; scope-specific programs must be qualified again.
A productive asset earns its place
in the next mission.
The independent references retain conventional wins and ties. Development acquisition costs match optimal conventional designs. A separate adaptive control is cheaper than the static compiler. This is a bounded software result, with commercial realization and broader operational autonomy still unproved.
03 / Earn the next capability
A larger possibility. A new burden of proof.
Demonstrated software
Three synthetic mission families. A preserved frozen experiment and a separately qualified current correction. Independent decision checks. A successor that inherits valid structure and refuses changed scope. Conventional ties and losses remain visible.
A physical diagnostic fixture
Construct a low-voltage switched network, reference loads and a calibrated acquisition interface. Blind the faults. Charge materials, bench time and labor. Test its decisions against an independent instrument.
Productive physical capacity
Qualified instruments can support manufacturing, commissioned infrastructure and useful energy services. Each stage needs its own materials, power, engineering, validation, financing and independently accepted demand.
The economic challenge remains open
At most US$100 total attributable cost → at least US$10,000 genuinely realizable incremental value (the original protocol used more than US$10,000). Status: UNPROVEN. There is no independent executable offer or cash receipt; full inference, compute, setup and labor costs remain unknown. Acquisition units are not dollars. Inspect the commercial test protocol.
04 / Inspect and reproduce
One experiment. A complete record.
Built additively on the existing SUCCESSOR Ω workbench. The experiment records its model, construction, source identity, typed work drafts and integrity commitments. Imported evidence remains inert.