The Hive Canon · The Complete Line Card

Everything we can prove.
One page, every primitive, all live.

AI has a trust gap. Models can explain what they say they did. They can't independently prove what they actually did. Hive exists to close that gap.

The proof is in the provenance. Provenance means signed, independently checkable evidence of what a model, agent, system, person, or machine actually did. This is the whole canon. It's every primitive we've built, what each one does, why a company would care, and the kind of team that lights up when they see it. Every row is a live page. Click through to see it working.

We can sign inference. We can sign audio and voice. We can sign what an agent did, and what a robot did. And we can anchor all of it so it verifies offline forever. It runs across the whole map, from frontier model labs and inference platforms to autonomous systems, wearables and health platforms, regulated exchanges, banks and custodians, and government and regulators.

If you know someone who should see this, the fastest thing in the world is a one-line intro. Look down the list, find the row that fits your people, and send them straight to the page.

Frontier model labsInference platformsAutonomous systemsWearables and health platformsRegulated exchangesBanks and custodiansGovernment and regulatorsMedia and content platformsData and ML platformsSecurity and infrastructure

These are buyer categories, not customers. Examples show where Hive could fit. They are not customers, partners, endorsements, or deployments.

Start here · plain words

A receipt is proof of what an AI actually did.

Not a claim it makes about itself. A signed, checkable record anyone can verify. We make it in three shapes.

Everything we can prove

Every proof we build, on one page. Scroll down to see the whole canon, each row a live page you can click into.

See the full canon ↓

Proof while it is streaming

Sign live voice, video, and agent streams as they happen, with no slowdown to the call. Built for HIPAA voice and call centers.

See Real-Time →

Proof for many agents at once

Run one AI agent or a hundred and get a receipt for every action, automatically. Pick the agent, the proof level, and the size.

The full canon

Every primitive, what it does, why they'd care, what we provide and why it pays, who it fits, and a click to the live page

The front door to the Canon Carnac™ Control Plane

Carnac™ sits before any provider or customer model and before any consequential tool, action, or effect. It selects, governs, and connects the proof instruments below. It does not replace the Canon. It conducts it.

Before inference CarnacPrompt™ commits the assembled prompt window before the model runs.

While you type Carnac Live Ink™ shows what needs context, proof, or permission without changing your words.

Before effect Carnac™ reads consequence and routes each read to the proof it warrants.

Challenge · escalate · stop Howler™ binds severity into a signed record and forces escalation.

Specialized primitives sit beneath and behind them. InkFrame v1, Proof Pre-Fill, Disclosure-Free Replay, Arrival Countersignature, SiGR, S2S, PPR, AFIR, MiR, Imprimatur, and the rest of the Canon produce the independent, cryptographically verifiable provenance.

Open the Carnac™ plane →

If a primitive has this symbol, it has a patent pending.

PrimitiveWhat it doesWhy they'd careWhat we provide · why it paysWho it fits

InkFrame v1 · The non-mutating substrate under Carnac Live Ink™

InkFrame v1Substrate · 8 content-addressed roots

Holds a proof-completion frame still while proof is attached to it, using 8 content-addressed roots and a hybrid Ed25519 + ML-DSA-65 signature over RFC 8785 JCS + SHA-256.

The physical evidence floor under the Carnac™ family. Any decision becomes an artifact that can be recomputed and reverified offline.

ProtectsNon-mutation guaranteed by math, not by policy. Change one byte and the address changes, which breaks the signature. Smoke suite 26/26, benchmark 5,000 frames / 0 integrity failures / 0 tamper escapes.

RFC 8785 JCSEd25519 + ML-DSA-65Post-quantum

Banks · regulators · audit

Proof Pre-FillFiling A · receipt before the act

The instant a decision forms, the receipt is prepared and bound by fingerprint only, before the action fires. Evidence or authority artifacts are resolved locally against a fingerprint, never raw text.

A ready receipt at the moment of decision. No lag, no after-the-fact reconstruction. The proof exists as the decision does.

PreventsThe "we'll write it up after" gap that lets machine decisions escape provenance.

Fingerprint-onlyPre-submit bindingNo raw text

Payments · treasury · agentic finance

InkFrame Non-MutationFiling B · content-addressed non-mutation

A content-addressed envelope sealed by its own hash. Change one byte and the address changes, which breaks the signature.

Tamper-evident record of what crossed. Anyone can recompute the hash to verify. Non-mutation is provable, not promised.

ProtectsEvery layer of the receipt from silent edits, by construction.

RFC 8785 JCSSHA-256Recompute-to-verify

EU AI Act Art 12 · NIST 800-53 AU-9/AU-10

Disclosure-Free ReplayFiling C · replay without exposure

Reconstruct the exact route and cue-delta of what happened without ever showing the prompt, the data, or the words. Built from a cue-delta manifest that references anchor_ids and fingerprints, not text.

Proof the route held without moving or exposing the data. Auditors and regulators see the shape of the reasoning, not the sensitive words.

ProtectsPHI, PII, trade secrets, and privileged content while still producing a full auditable trail.

Cue-delta manifestAnchor-ids onlyLeak detection built-in

Healthcare · legal · finance

Arrival CountersignatureFiling D · gateway arrival check

The gateway counter-signs on arrival by comparing the approved action tuple against the delivered action tuple. If they match, the gateway signs. If not, the delta is recorded and the action refused.

Proof of the journey, end to end. What ran matches what was cleared, and that match is itself a signed artifact.

PreventsSilent substitution of recipient, amount, or target between clearance and execution.

Approved vs deliveredSigned matchSigned refusal

Wires · payouts · agent-to-agent commerce

Upstream · pre-effect receipts Filed · USPTO 64/119,279 · Jul 26, 2026

PBSProvenance-Bonded Sandbox

Binds container, kernel modules, package index, egress ACL, GPU firmware, and every runtime measurement into a Merkle heartbeat accumulator, chained across the life of the environment.

Confidential-computing attestation runs once at boot. Everything after is invisible. PBS makes silent alteration of the sandbox mathematically impossible to hide.

ProtectsContinuous provenance of the execution environment. A kernel-module swap, in-memory package replacement, or dropped heartbeat all break the chain deterministically.

ML-DSA-65Heartbeat Merkle chainOffline-verifiable

Frontier labs · regulated inference

Refusal LedgerPublic mutation ledger · ZK envelope-bond

A chained Merkle mutation ledger of every refusal-policy change, plus a Pedersen commitment and zk-SNARK envelope-bond that proves the runtime policy value falls inside a public bound without disclosing it.

Refusal thresholds moved silently at OpenAI. That fact had no external witness. Refusal Ledger makes every mutation externally verifiable and lets the operator prove lawful bounds without revealing the number.

ProtectsExternal verifiability of policy state at inference time. Operators keep the value confidential. Regulators, insurers, and courts verify the envelope.

Pedersen commitmentszk-SNARK envelope-bondCT-log-style ledger

Frontier labs · EU AI Act Article 12

HowlerSAE-triggered signed freeze receipt

Sparse-autoencoder feature indices, magnitudes, and a deterministic vector digest signed at the moment an interpretability probe fires inside the reasoning loop. Three variants: drift, capability, contamination.

Alarm-about-intent is normally producer-controlled. Howler makes it independently replayable: a third party can re-run the same trace prefix through the same probe and confirm the flagged features fired at the recorded magnitudes.

ProtectsExternally reproducible evidence of dangerous-intent detection. Extends the existing "challenge, escalate, stop" primitive with the patent-pending SAE-triggered variant.

ML-DSA-65SAE feature commitDeterministic replay

Model developers · safety orgs

Perimeter BondEnforcement bytecode bonded to every attempt

Every egress-attempt receipt embeds ebpf_program_hash = SHA-256(compiled_ebpf_bytecode). The manifest declares the same hash. Any silent swap of enforcement code between manifest and interception is externally visible.

Existing egress records say "we blocked X." Nobody outside can verify the bytecode doing the blocking wasn't swapped. Perimeter Bond makes silent enforcement substitution externally detectable.

ProtectsIntegrity of the enforcement code itself. Attempt-time hash must match manifest-time hash. Mismatch is the alarm.

eBPF bytecode hashML-DSA-65Manifest ↔ attempt bond

Inference platforms · regulated egress

Diurnal BondDynamic k-of-n for weekend, on-call, incident

Consequential actions during weekend, on-call, after-hours, or declared-incident windows require k-of-n countersigning from geographically distributed attestors. k is dynamically computed from a signed risk manifold, indexed by (regime, action class), with the manifold digest bound into every regime-declaration receipt.

The Hugging Face incident compounded over a weekend. Nobody was awake to countersign. Diurnal Bond forces the required humans to be countersigning, geographically, before the weekend action can commit.

PreventsSilent autonomous action during high-risk temporal regimes. The clock cannot be gamed; the manifold digest is signed.

Threshold signaturesSigned risk manifoldGeo-distributed attestors

Frontier labs · agentic finance · ops

Egress BondSemantic-class exfil meter · retroactive DAG invalidation

Additive Pedersen commitments meter exfiltration per semantic class (credential, PII, model weight, test data, plaintext) against pre-committed caps, using a Merkle-committed local classifier. Cap breach breaks the bond and can retroactively invalidate the associated SiGR Chain DAG.

17,000 test-data rows walked out over a weekend. Reconstruction was archaeological. Egress Bond breaks the bond at row 10,001 and retroactively invalidates the agent DAG. Reconstruction becomes cryptographically unnecessary.

PreventsSilent per-class exfiltration past a signed cap. The producer never sees which specific rows contributed; only per-class totals.

Pedersen homomorphismMerkle-committed classifierRetroactive DAG invalidation

Dataset platforms · regulated data flows

Forensic RailConsortium-bonded, replay-bonded incident analysis

Post-incident analysis runs under a threshold-signed k-of-n consortium credential (ISAC-style). Every model call under that credential is deterministic-replay-bonded: seed, temperature=0, top_p, and prompt/output/KV-cache digests all embedded so a third party can byte-identically reproduce the output.

After the HF incident, forensic analysis ran on the producer's own account. No independent responder could replay it. Forensic Rail makes the incident response itself receipt-bonded and reproducible.

ProtectsIndependent reproducibility of the responder's analysis. Removes the ability to publish work that a regulator or court cannot replay.

Threshold-signed credentialDeterministic replayISAC-style consortium

Incident responders · regulators · auditors

Inference & model proof

SiGRSigned Inference Guarantee Receipt

Signs the decision itself, meaning the model, inputs, and outcome, into one ML-DSA-65 receipt anchored on Base.

The atom of the whole stack. Any AI output becomes independently verifiable without trusting the vendor.

ProtectsIndependent evidence a decision happened as claimed. It turns "trust our logs" into a signed exhibit that stands up in a dispute or audit.

ML-DSA-65Offline-verifiableSOC 2 / MRM evidence

Frontier model labs

S2SSilicon-to-Signature

Binds each inference into a hash-chained attestation epoch anchored to the hardware evidence and a Hive-issued nonce, signed with ML-DSA-65.

Confidential-computing attests the GPU once at startup, but every inference after that is unbound. S2S closes that gap. It proves what ran, where, under what model, and in what unbroken order.

ProtectsThe compute-side root of trust. It proves occurrence and custody of a run without trusting NVIDIA, the provider, the model owner, or Hive.

ML-DSA-65Zero-network verifierCrypto core 11/11 · real silicon pending

Inference platforms

PPRPhysiological Provenance Receipt

Signs the claim about a body, meaning that a reading came from a real on-body sensor, on the real person, in an unspliced stream, into an ML-DSA-65 receipt. The raw signal never touches Hive.

Any product that turns human telemetry into an AI claim, like readiness scores, AFib flags, RPM alerts, or clinical endpoints, can prove the reading was real. Anyone can check it offline: a regulator, a court, or the other side's expert.

ProtectsProvenance of a physiological claim without ever ingesting the raw signal ("the wall"). Does not diagnose; on-body presence is discontinuity detection, not biometric identity.

ML-DSA-65Signal stays device-sideOffline-verifiable

Wearables and health platformsRegulated exchanges

AFIRAttested Fragmented Inference Routing

Breaks a request into signed, routable sub-tasks and signs each fragment inside the inference path.

Proof at the fragment level for agentic pipelines and huge documents. You get speed, cost savings, and an audit trail all at once.

SavesProof without re-running the whole job. You only sign the fragments that matter, so audit cost stays flat while document size and token spend don't balloon.

O(1) selective attestationML-DSA-65Lower audit compute

Inference platforms

MiRModel Identity & Relineage

Signs which model actually served each step of a run, and detects substitution against the model you contracted for.

Answers model-risk-management: was this the model you approved, and did it stay that model for the whole run.

ProtectsPins every step of a run to the exact model version that served it. This is the evidence model-risk and EU AI Act reviewers demand, so a bad output can't be blamed on a mystery model.

MRM evidenceEU AI Act readyML-DSA-65

Frontier model labsData and ML platformsInference platforms

HiveBoundInput-side pre-commitment

Turns anonymous calls into signed envelopes declaring identity, intent, and budget before inference runs.

Proof starts at the input, not the output. The request is accountable from the first byte.

SavesKills unaccountable spend at the door. Identity, intent, and budget are signed before a token is burned, so runaway or abusive calls never bill.

Fail-closed gateBudget-boundML-DSA-65

Frontier model labsSecurity and infrastructureInference platforms

ImprimaturPre-attestation gate

Signs a clearance before an inference runs and refuses any call without a valid, unexpired pass.

A policy checkpoint that proves the rule was enforced up front, not reconstructed after the fact.

ProtectsProves the policy fired before the call ran. It's a signed pre-clearance, not a reconstructed after-the-fact story, so "the control was enforced" holds up to a regulator.

SOC 2 control evidenceFail-closed gateML-DSA-65

Security and infrastructureData and ML platformsFrontier model labs

XCALIBURVerifiable inference suite

Public trust-receipt + benchmark surface that proves AI decisions rather than asserting them.

The "show me" bar for any model vendor claiming quality. Signed benchmarks beat marketing.

Makes moneyTurns a quality claim into a sale. Signed benchmarks a buyer can re-verify shorten the "prove it" cycle and win deals that marketing numbers lose.

Public verify embedRe-verifiableML-DSA-65

Frontier model labsData and ML platforms

Audio, voice & media

AFiR-StreamReal-time voice & streaming

Signs streaming segments off the hot path and seals a post-quantum session anchor at the close.

Live voice and audio become provable. Every second of a call or stream is receiptable, with no latency hit.

ProtectsMakes a call or stream defensible after the fact. Every segment is receiptable with zero latency hit, so consent, disclosure, and "what was actually said" survive a dispute.

Zero added latencyPost-quantum session anchorML-DSA-65

Media and content platforms

Media Origin ReceiptM.O.R.

Signs media origin and rights across human, AI, hybrid, and unknown provenance states.

Provenance for the content itself. Who or what made this, and can it be licensed.

Makes moneyMakes content licensable and defensible. Signed origin and rights across human, AI, and hybrid sources unlock clean licensing revenue and cut off downstream IP claims.

Licensable rightsC2PA-compatibleML-DSA-65

Media and content platformsFrontier model labs

MoRSo · Segments of OriginMoRSo

The receipt every clip, quote, and excerpt carries. It builds on MoR to make meaning-altering cuts structurally visible, while honest edits still verify clean.

The subject can refute a distorted clip with proof, not a denial: "91% of the answer was omitted, the question was removed." This only checks fidelity to the source. Framing is never assessed.

ProtectsTurns a viral out-of-context clip into a provable, refutable claim. Cutting, splicing, and reordering across sealed boundaries show up as meaning-bearing changes, while a simple cough cut still passes clean. Hive never receives the media.

Sealed boundary mapNon-revealing commitmentsML-DSA-65

Media and content platformsFrontier model labs

Signed Royalty-ProvenanceSRPR

Binds receipts to royalty, canon, and media-rights workflows for downstream settlement.

Ties provenance to money. Proof that flows straight into royalty accounting.

Makes moneyGets the right party paid, provably. Provenance ties straight into royalty accounting, cutting disputed splits and leakage on every payout.

Settlement-linkedCuts payout disputesML-DSA-65

Media and content platforms

OriginProofHuman-Origin Attestation

Binds the conditions under which expert human work was produced: credentials, session, tools, and output.

Proves data was made by real qualified humans. It's the trust layer for expert-data marketplaces.

Makes moneyLets you charge a premium for verified-human data. Signed credentials, session, and tools prove the work wasn't AI-farmed, which commands higher rates and passes buyer audits.

Credential-boundPremium data tierML-DSA-65

Data and ML platforms

AFiR-OCR DocProofDocument extraction

Proves what a document model read, where on the page, and at what confidence.

Extraction you can defend. The OCR result is signed to its source region.

ProtectsDefends every extracted number to its source. The value, page region, and confidence are all signed, so an OCR error in a contract or filing is provable and traceable, not deniable.

Source-region boundEvidentiary trailML-DSA-65

Frontier model labsSecurity and infrastructureBanks and custodians

Agents & autonomy

AFiR-S3Agentic receipts

Signs tool scope, action, context mutation, and reward around every autonomous-agent step.

Turns an agent workflow into an audit trail. Every step is provable.

ProtectsBounds what an agent can do, and proves it did only that. Signing before acting on scope and action limits liability when an autonomous agent touches customer systems.

Fail-closed gateLiability boundaryML-DSA-65

Data and ML platformsSecurity and infrastructureFrontier model labs

SmartAgent Route GraphRoute proof

Shows an agent moving across model / provider / tool / payment stops, each stop emitting a receipt.

See and prove the whole trajectory. Route proof, not just a final answer.

ProtectsShows exactly where a run went wrong. A receipt at every model, tool, and payment stop pinpoints the failing hop, so blame and remediation land on the right vendor.

Full-trajectory proofFault isolationML-DSA-65

Data and ML platformsFrontier model labsSecurity and infrastructure

Agent Trip & SPIREInstance + trajectory attestation

Treats a run as a trip and signs agent-instance mints, trajectories, and coherence as FIPS 204 receipts.

Identity and coherence for agent fleets. Every agent instance is attested.

ProtectsProves which agent instance did the work. SPIRE-attested identity plus trajectory receipts stop rogue or spoofed agents from acting unattributed in a fleet.

SPIRE attestationAttributable identityFIPS 204

Security and infrastructureInference platforms

HiveGate & HiveMorphOnboarding + discovery

Mints agent identity, translates intents across runtimes, and exposes Hive's tool index and trust scoring.

The on-ramp and lookup for cross-platform agents. One identity that travels, discoverable by agents.

Makes moneyGets your agents found and trusted to transact. One portable signed identity plus a trust score makes agents discoverable across runtimes, opening agent-to-agent revenue.

MCP-discoverablePortable DIDTrust score

Security and infrastructureData and ML platformsFrontier model labs

Robotics & physical autonomy

Hive Provable MachinesRobotics / RaaS

Applies receipts, proof vectors, and evidence exports to robot task handoffs, uptime, and teleop.

Independent proof of what a machine actually did. The audit layer for autonomy.

ProtectsIndependent proof of what a machine did, for the incident file. Signed task-handoff and uptime receipts export straight into safety, insurance, and litigation defense.

Insurance evidenceIncident exportML-DSA-65

Autonomous systemsInference platforms

Protected Flow FleetsFleet proof container

Organizes many protected flows by business unit, site, or robot fleet into one proof-state view.

Fleet-scale evidence rollup. Prove a whole fleet's behavior, not just one unit.

SavesOne evidence view for a whole fleet. Roll up proof across every unit and site so regulatory and insurer reporting is one signed export, not a per-robot scramble.

Fleet rollupOne-export reportingML-DSA-65

Autonomous systems

Trust vectors, settlement & economics

R3PvReceipt Proof Vector

Groups receipts into a signed decision vector: verification depth, weakest boundary, recoverability, next action.

One machine-readable trust score for a whole event. The weakest-link finder.

SavesFinds the weakest link before it costs you. One signed score per event flags the thinnest boundary and whether it's recoverable, so risk teams triage instead of investigating everything.

Weakest-link scoreFaster triageML-DSA-65

Security and infrastructureBanks and custodians

Protected FlowRisk & recovery layer

Reframes payment receipts as risk, compliance, and recovery infrastructure for high-value flows.

Turns a receipt into a recoverable, insurable event. Proof with money attached.

SavesMakes a bad payment recoverable and insurable. Risk and recovery receipts on high-value flows cut chargeback and fraud loss, and give insurers a signed basis to cover it.

Recoverable / insurableCuts fraud lossML-DSA-65

Banks and custodians

Hive Receiptx402 / Base USDC settlement

Spectral-signed Base USDC payment receipts with on-chain verification and an MCP endpoint.

The settlement rail. Every agent payment is a verifiable on-chain receipt.

Makes moneyLets agents pay per call, provably. On-chain USDC receipts settled via x402 open up sub-cent metered revenue, with an on-chain audit trail and an MCP endpoint.

Base USDC / x402Per-call revenueOn-chain verify

Banks and custodiansSecurity and infrastructure

Proof CreditEconomic-evidence primitive

Converts machine-readable proof-state into underwriting-grade risk signals and bundles.

Proof becomes a credit signal. The bridge from receipts to underwriting.

Makes moneyTurns proof into a priced risk signal. Underwriting-grade proof-state profiles let insurers and lenders quote against signed evidence, unlocking new products.

Underwriting-gradeNew risk productsML-DSA-65

Banks and custodians

Structural LaterationResidual-cost metering

Lets a new attested shape laterate from prior shapes and pay only the residual cost.

Metering that rewards reuse. You pay for what's new, not what's already proven.

SavesYou pay only for what's actually new. Residual metering credits proof reused from prior shapes, cutting attestation and compute spend on repeated work.

Residual meteringPay-for-new-onlyML-DSA-65

Inference platforms

Hive Settlement FeedMarket-data integrity

Independently signs the data values that regulated markets settle against.

Provable settlement data. The number the market pays out on is signed.

ProtectsRemoves the "who set the number" fight. Independently signed settlement values are the payout basis, so a regulated market can defend every settlement it makes.

Independent data-integrityDefensible payoutML-DSA-65

Regulated exchanges

Hardware & sovereign trust

Hive Typed SignerML-DSA-65 engine

The shared typed-fragment signing engine underneath SiGR, AFiR, MiR, GCA, and the rest.

The core cryptographic engine. Everything above runs on it.

SavesPost-quantum signing you don't have to build. One FIPS 204 sign-and-verify API drops into any workflow, saving you a from-scratch crypto migration.

FIPS 204 / ML-DSA-65Drop-in APIEd25519 fallback

Inference platformsSecurity and infrastructureGovernment and regulatorsFrontier model labs

HiveSeal · QPuFHardware root of trust

A signing device that keeps the key in silicon, born from QRNG entropy and bound to the device.

Keys that are random at birth and un-stealable for life. Trust that lives below the software.

ProtectsA stolen key can't forge your receipts. QRNG-born, PUF-bound keys sign in silicon and never leave the device, closing off the biggest signing-trust attack.

QRNG + PUFUn-exportable keyIn-silicon ML-DSA-65

Inference platformsSecurity and infrastructureFrontier model labs

Sovereign AI Receipt RegistryCountry-scale proof

Every high-risk national AI decision signed at the source with sovereign-held keys.

A country owns its own proof layer. Verifiable past the quantum era.

ProtectsA country owns its own proof, with no foreign dependency. High-risk national AI decisions are signed at the source with self-held keys, durable and auditable past the quantum era.

Sovereign key custodyPost-quantum durableML-DSA-65

Government and regulators

Public proof surfaces

Hive Command CenterLive proof dashboard

Public live receipt streams, trust-network canvas, and compliance status.

The "it's real, watch it" surface. Receipts flowing in public, right now.

Makes moneyShows a skeptical buyer it's real, live. A public feed of receipts flowing right now shortens the trust conversation and moves deals past "prove it works."

Live receipt streamsShortens sales cycleCompliance status

Frontier model labsSecurity and infrastructureAny buyer

Hive Customer ConsoleTenant dashboard

Customer-facing receipts, spend, keys, and evidence bundles, and it works the same no matter which primitive you use.

The self-serve product surface. A buyer runs their own proof from day one.

SavesAudit packs on demand, no ticket needed. Self-serve receipts, keys, and evidence bundles export a SOC 2 / audit-ready pack in a click instead of a quarter-long evidence hunt.

Evidence bundlesSOC 2 exportSelf-serve

Data and ML platformsAny buyer

HVVSVerifiable Voting System

Cryptographically verifiable receipts for electronic ballot-return workflows.

Election integrity as independent proof. The audit layer for democracy.

ProtectsMakes an election defensible after certification. Signed proof cards and audit records outlive the event, so a contested result can be independently re-verified, not re-argued.

Post-quantum durableIndependent recountML-DSA-65

Government and regulators

Send it to anyone.

This whole page is shareable, and every row goes to a working product. We can do inference, audio, agents, robots, settlement, and hardware. Find the row that fits your people and send them straight to it. A one-line intro moves faster than anything else.