/* Variation A — Editorial / Institutional
Inspired by FT / Bloomberg infra-vendor pages. Dark navy hero,
restrained typography, left-aligned with side meta. The auditor
reads this and feels like a serious counterparty.
*/
// Reads a body data-* attribute and re-renders when Tweaks change it.
// Falls back to `def` if unset. Used so the same components work both
// inside the Landing Page.html compare shell and the standalone single-page.
const useBodyAttr = (key, def) => {
const read = () => typeof document !== 'undefined' && document.body.dataset[key] || def;
const [v, setV] = React.useState(read);
React.useEffect(() => {
const sync = () => setV(read());
sync();
window.addEventListener('tweakchange', sync);
return () => window.removeEventListener('tweakchange', sync);
// eslint-disable-next-line react-hooks/exhaustive-deps
}, [key, def]);
return v;
};
const VariationABrand = ({ inv }) =>
R/001RegistrumAI;
const VariationANav = () => {
const style = useBodyAttr('navStyle', 'solid');
const [open, setOpen] = React.useState(false);
// Close the menu after navigating, and whenever the viewport grows back to desktop.
React.useEffect(() => {
if (!open) return;
const onResize = () => { if (window.innerWidth > 760) setOpen(false); };
window.addEventListener('resize', onResize);
return () => window.removeEventListener('resize', onResize);
}, [open]);
const links = [
{ href: '#governance', label: 'The Gap' },
{ href: '#how', label: 'How it Works' },
{ href: '#product', label: 'What it Does' },
{ href: '#compliance', label: 'Learn More' },
];
return (
);
};
const VariationAHero = () => {
const headline = useBodyAttr('heroHeadline', 'every-call');
return (
Enterprise AI Agent Governance
{headline === 'three-signatures' ? <>Three signatures, one ledger.> :
headline === 'govern-audit' ? <>Govern every agent. Audit every call.> :
<>Every agent. Every call. On the record.>}
RegistrumAI is the governance plane between your AI agents and every LLM provider — real-time policy enforcement, a cryptographic audit trail, and buyer-side cost controls, in one drop-in layer your CISO, CFO, and Procurement team can all approve. No SDK. No instrumentation. Sub-5ms.
CoverageOpenAI · Anthropic · Google · Groq · Mistral
DeploymentSelf-hosted · SaaS
✓ SOC 2 in progress✓ ISO 27001 roadmap✓ HIPAA-ready✓ EU AI Act aligned
);
};
const VariationAProblem = () =>
§ 01 · The governance gap
Your auditors will ask four questions. Most enterprises can't answer them.
AI agents are writing code, processing invoices, and routing customer decisions across your organization. Security, finance, and compliance are forming their questions now.
{[
['01', 'What agents exist?', 'Discover and inventory every AI agent operating in your network — including the ones you don\'t know about. Shadow AI surfaces in minutes, not quarterly audits.'],
['02', 'What are they spending?', 'Real-time cost metering at the token level, attributed to a cost center. Chargeback reports use your contracted rates, not published rack pricing.'],
['03', 'Are they authorized?', 'Policies enforced at the proxy layer — model allowlists, data residency, DLP, rate limits. The auditor sees the policy and the log of every disposition.'],
['04', 'What did they do?', 'A tamper-evident, hash-chained audit trail of every call. Full prompt and response capture available with PII masking and retention controls.']].
map(([n, t, b]) =>
Q.{n}
{t}
{b}
)}
;
const VariationAHow = () =>
§ 02 · How it works
A drop-in proxy between your agents and every LLM.
01
Point your agents at one endpoint.
Change a base URL. No SDK, no library, no code modifications. Any agent that speaks OpenAI speaks Registrum.
02
Identify, price, and govern every call.
API key fingerprinting names the agent. Your contracted rates price the tokens. Your policy decides whether to forward, block, or redact — before the call leaves your network.
03
Hand the auditor the ledger.
An immutable, hash-chained log of every call — agent, model, tokens, cost, disposition. Pre-built evidence packs for SOC 2, ISO 27001, HIPAA, and the EU AI Act.
;
const VariationAFeatures = () => {
const features = [
{ tag: 'Discovery', t: 'Shadow AI surfacing', d: 'Every unregistered agent on your network appears in the fleet view from its first call, with full history and attribution.' },
{ tag: 'Metering', t: 'Contracted-rate pricing', d: 'A customer-owned pricing catalog using your negotiated rates. Token-level cost attribution to agent, team, and cost center.' },
{ tag: 'Policy', t: 'Pre-execution enforcement', d: 'Model allowlists, data residency, DLP, rate limits, time-of-day controls — applied at the proxy before the call leaves the network.' },
{ tag: 'Audit', t: 'Hash-chained ledger', d: 'A tamper-evident record of every call. Optional full prompt/response capture with PII masking and retention policies.' },
{ tag: 'Security', t: 'DLP & secret detection', d: 'Outbound prompts inspected for PII, source code, credentials, and prompt-injection patterns. Block, redact, or alert per policy.' },
{ tag: 'Platform', t: 'GitOps-native', d: 'Terraform provider, REST API, CI/CD gates, SIEM and SOAR integrations. Agent governance ships in the same PR as the agent.' }];
return (
§ 03 · What it does
One control plane. Six surfaces.
{features.map((f, i) =>
{f.tag}
{f.t}
{f.d}
)}
);
};
const VariationABuyers = () => {
const buyers = [
{
role: 'CISO', mark: 'S', label: 'CISO / Head of Security',
h: 'Prove to your auditor — not your vendor — what every agent did.',
p: 'Tamper-evident audit evidence your security team can own, with keys you hold — not keys your LLM vendor manages.',
proofs: [
'BYOK: your keys, your audit chain',
'OPA-enforced policy at the call boundary',
'Tamper-evident HMAC manifests, buyer-side',
'Pre-mapped to SOC 2, ISO 27001, EU AI Act'],
cta: 'Get the CISO brief', rid: 'brief_ciso',
doc: 'briefs/ciso-brief.html'
},
{
role: 'CFO', mark: 'F', label: 'CFO / FinOps',
h: 'Forecast agent spend the way you forecast cloud spend.',
p: 'Your own meter — not the vendor\'s invoice. Every LLM token and outbound action, classified and reconciled against your approved budgets.',
proofs: [
'Per-call telemetry: LLM tokens and tool actions',
'Vendor-meter reconciliation — variance, not vibes',
'Cost envelopes by team, workflow, agent identity',
'Hard budget ceilings with enforcement, not alerts'],
cta: 'Get the CFO brief', rid: 'brief_cfo',
doc: 'briefs/cfo-brief.html'
},
{
role: 'PROC', mark: 'P', label: 'Procurement / Vendor Management',
h: 'Walk into the renewal with your meter, not theirs.',
p: 'Buyer-side evidence designed to sit on the table at the renewal meeting — exportable, vendor-reconciled, with a contract-clause library to match.',
proofs: [
'Renewal Cost-Envelope Report, procurement-ready',
'Vendor Agent Licensing Audit Checklist',
'Contract-clause library: meter, audit, dispute',
'Complementary to vendor agents — leverage, not conflict'],
cta: 'Get the Procurement brief', rid: 'brief_proc',
doc: 'briefs/procurement-brief.html'
}];
return (
§ 04 · Three signatures
Built for everyone who has to say yes.
Enterprise AI deployment requires sign-off from security, finance, and procurement. Each stakeholder gets the evidence they need to approve — and to stay approved at renewal.