T Tanilo
For Business

Content AI optimizes what agents read. Tanilo verifies what agents do.

Enterprise content platforms make your content easier for AI agents to consume. Tanilo does the opposite: we verify the claims your AI agents are about to act on, before they act. Pre-action fact-checking with a cryptographic receipt your legal team can audit. Anyone can verify it offline in Node, Python, or the browser — no Tanilo service required.

Start a pilot conversation → See the developer side
who buys this

Three teams already feel this pain.

If your team publishes AI-generated content at any scale, one wrong claim is a brand crisis or a compliance violation. Tanilo is the pre-publication checkpoint with a signed audit trail.

Content Agencies
Pain: one bad campaign claim = client loss

Generate at scale, verify before send. Hand the brand a signed audit trail proving every claim was checked. Pricing scales with your output, not their team size.

Regulated Industries
Pain: EU AI Act, NIST RMF, SEC, FDA disclosures

Financial services, healthcare, insurance, pharma. AI-generated copy with regulatory implications needs a cryptographic record proving verification before publication. Receipts are tamper-evident and replayable for audit.

Enterprise Brand Teams
Pain: AI claim shows up in a press release

Plug Tanilo into your approval workflow. Marketing draft → verification check → receipt attached → legal sign-off. No manual fact-check team. No screenshot evidence. Just a signed JWS.

how it works

Five steps from claim to signed audit trail.

01
Submit

Send a claim or document to /evaluate via API or MCP.

02
Verify

Multiple quorum members check the claim in parallel, plus an adversarial re-check pass, with sources retrieved per claim.

03
Score

Per-claim confidence with sources and adversarial test result.

04
Sign

Response signed with our Ed25519 key. Public JWKS for verifiers.

05
Audit

Store the receipt. Replay the verification any time. Hand to legal.

pricing

Four tiers. Plain.

No procurement gymnastics, no annual contracts, no per-seat counting. Pay for what you verify.

Developer
$99 / month

Self-serve is launching — email us for manual setup in the meantime. 2,000 verifications a month, signed receipts, cancel anytime. Priced for individual builders and small teams.

  • 2,000 verifications / month included
  • Launching — email for manual setup today
  • Same signed receipts as enterprise tiers
  • Verify offline: pip install agentoracle-receipt-verify
  • Public JWKS, public spec, public benchmark
See self-serve →
Business Pilot
$1,000 / 60 days, money-back

60-day money-back pilot. Prove the audit trail on your real content before committing to a monthly line item.

  • 10,000 verifications total ($0.10 each)
  • Custom dashboard with audit log export
  • Async Slack/email support
  • Receipt format customization
  • Priority queue access — current measured latency 9–14s per request
  • Money-back if you cancel before day 60
Start a pilot →
Continuation
$2,500 / month

Ongoing monthly plan after the pilot. Same $0.10 per verification, no annual lock-in, cancel any month.

  • 25,000 verifications / month included
  • Overage at $0.10 / verification
  • Custom dashboard + audit log export
  • Async Slack/email support
  • Priority queue access — current measured latency 9–14s per request
  • Month-to-month, cancel any time
Talk to us →
Enterprise
Custom

High volume, custom SLA, dedicated signing key (kid) for receipts, on-prem deployment options.

  • Unlimited verifications
  • Dedicated signing key (your kid value in JWKS)
  • SLA guarantees and 99.9%+ uptime
  • Compliance review support (EU AI Act Article 12, NIST AI RMF, SOC 2 on roadmap)
  • Annual contract, MSA, security review
  • Engineering response time guarantees
Talk to us →
why we are different

Cryptographic proof, not vendor claims.

IETF Internet-Draft — Filed draft-krausz-verification-state-02 Filed June 6, 2026 · -02 published September 23, 2026 · 36 pages · individual submission on the IETF datatracker · 10 RFC references · sibling to environment-state family. View on datatracker → Public Receipt Spec v0.3 agentoracle-receipt-spec Binary-halt gate · canonical/derived/version-bound mapping · 2 ADRs · content-addressed via v_gate_mapping_hash · MIT licensed. View on GitHub → Reference Verifier agentoracle-receipt-verify TypeScript JWS verifier · offline verification of any v0.3 receipt · no facilitator dependency · MIT licensed. View on GitHub → AVeriTeC 2024 Benchmark agentoracle-benchmark v0.1 57.6% overall · 57.7% held-out (vs ~30% paper baseline) · verdict mapping selected by inspection on the calibration half · retrieval recall not reported · contamination not controlled for · internal baseline, not a leaderboard result · open methodology · open submissions · MIT licensed. View results →

Built for the regulations coming next

  • EU AI Act Article 12. Record-keeping for high-risk systems, applicable from 2 December 2027 for standalone Annex III systems and 2 August 2028 for AI embedded in Annex I products, under Regulation (EU) 2026/1744. Cryptographic receipts are tamper-evident and replayable.
  • NIST AI RMF Measurement. Per-claim confidence + provenance + source attribution.
  • SOC 2 Type II. On roadmap, scoped with the Enterprise tier rollout.
  • Public JWKS. Any auditor can verify our receipts without trusting us. RFC 7515 / 7517 / 8037.

Ready to talk?

Tell us your use case in 3 sentences. We’ll come back same-day with a yes/no on fit, a sample receipt against your content, and a pilot scope.

joe@tanilo.io