Current-state disclosure · not a certification
Missing evidence is labeled missing; planned controls are not presented as deployed controls.
No production p99 authorization-gate latency or capacity SLA is published yet. Browser-demo microseconds are local simulation timing and must not be used for production sizing. A pilot must pre-register server-side p50/p95/p99, timeout, and saturation measurements.
BrainSignal begins in shadow mode. A mandatory fail-closed gate can stop spend, so enforcement requires a buyer-approved availability target, timeout behavior, queued/manual exception path, break-glass ownership, and auditable bypass policy. No universal degraded-mode policy is claimed today.
Named rails describe categories the authority model can govern; they are not certified connector claims. The public demonstration uses /v1/sentinel/evaluate. Production webhook, authentication, connector, and ERP contracts remain buyer- and rail-specific until published.
An authorization receipt can bind the proposed action and later rail reference, but final settlement, return, reversal, and dispute status require rail-specific ingestion. No complete multi-rail settlement adapter matrix is claimed. A pilot must name the authoritative outcome feed and reconciliation owner.
The public receipt specification, refusal demo, and open package are inspectable. Independent multi-party custody, external anchoring, key rotation, revocation distribution, and recovery exercises remain operational validation work.
Public source has a security policy. BrainSignal does not currently claim SOC 2 certification, a completed independent penetration test, universal PCI scope exclusion, a published DPA/subprocessor schedule, or committed data-residency regions.
No fixed BrainSignal price is published. Initial work is a scoped design-partner pilot; commercial terms depend on workflow, evidence retention, integrations, support, and deployment boundary. Titan Guard is separately available through AWS Marketplace.
No named customer deployment or production fraud-loss result is claimed. Current public results are internal synthetic evidence. Customer outcomes will be added only after a pre-registered evaluation and buyer-approved disclosure.
Patent counts are not used as buyer-facing product evidence because a public patent schedule has not been published. Startup credit and enablement programs are labeled as programs used—not customer, channel, integration, or validation partnerships.
Card controls, JIT funding, spend platforms, and AP suites are strong within their own rail or workflow. Agent-payment protocols increasingly standardize identity and mandate expression. BrainSignal’s intended differentiation is narrower: one pre-execution authority decision across heterogeneous rails, plus a portable receipt that can be verified outside the originating payment or procurement system. A buyer should adopt BrainSignal only if that cross-rail policy and evidence portability adds value beyond controls already owned.