# THRONDAR

> THRONDAR is a governed multi-model AI council. A council answer is normally synthesised from several frontier models (the signed receipt records when a single-model fallback answered instead). In the signed-in console its receipt is signed with ML-DSA-87 (FIPS 204), and the model-plane bridge co-signs it with Falcon-1024 (FN-DSA, pre-standard: NIST has not yet published FIPS 206 and has announced that its signing procedure will differ), so its origin and integrity can be checked offline; a signature never attests that an answer is correct. Studio adds metered image, voice and video generation; Autopilot runs bounded multi-step council runs sealed with a signed receipt.

## Product

- [Studio](https://throndar.ai/studio): metered image, voice and video generation; FLUX is live by default, other engines are switched on per deployment
- [Autopilot](https://throndar.ai/autopilot): bounded council runs (at most 8 steps) sealed with an ML-DSA-87 receipt
- [Home](https://throndar.ai)
- [Pricing and plans](https://throndar.ai/pricing)
- [The council: which models answer](https://throndar.ai/council)
- [Compare](https://throndar.ai/compare)
- [Forge](https://throndar.ai/forge)
- [Security](https://throndar.ai/security)
- [Standards](https://throndar.ai/standards)
- [Connectors](https://throndar.ai/connectors)
- [Showcase](https://throndar.ai/showcase)
- [Frequently asked questions](https://throndar.ai/faq)
- [Redeem a code](https://throndar.ai/redeem)
- [Status](https://throndar.ai/status)
- [Privacy policy](https://throndar.ai/privacy)
- [Terms of service](https://throndar.ai/terms)
- [Verify a proof](https://throndar.ai/verify/proof)
- [Verify offline](https://throndar.ai/verify/offline)
- [Support](https://throndar.ai/support)
- [Post-quantum cryptography](https://throndar.ai/pqc)
- [Cryptographic Bill of Materials](https://throndar.ai/cbom)
- [Evidence](https://throndar.ai/evidence)
- [Governance](https://throndar.ai/governance)
- [AI readiness](https://throndar.ai/ai-readiness)
- [EU AI Act](https://throndar.ai/ai-act)
- [Developers](https://throndar.ai/developers)
- [Learn](https://throndar.ai/learn)
- [Transparency ledger](https://throndar.ai/transparency/ledger)

## Learn

- [What “verified AI” actually means — and what it doesn’t](https://throndar.ai/learn/what-verified-ai-means): A cryptographic signature on an AI answer proves where it came from and that it wasn’t changed — not that it’s true.
- [Why one AI model is a single point of failure](https://throndar.ai/learn/single-model-single-point-of-failure): One model is one training run and one set of blind spots. A council makes disagreement a signal you can see.
- [The NIST post-quantum cryptography standards, explained](https://throndar.ai/learn/nist-post-quantum-standards-explained): Three finalized standards (FIPS 203/204/205), one forthcoming (Falcon/FIPS 206), and why the transition is a multi-year job.
- [AI provenance and the EU AI Act: the evidence you actually need](https://throndar.ai/learn/ai-provenance-eu-ai-act-evidence): Signed outputs, an append-only log, and a crypto inventory produce evidence that supports your obligations — not a certification.
- [What is a Cryptographic Bill of Materials (CBOM)?](https://throndar.ai/learn/what-is-a-cryptographic-bill-of-materials): A machine-readable inventory of the crypto you actually use — and the honest first step of any post-quantum migration.
- [How to verify an AI answer: a practical guide](https://throndar.ai/learn/how-to-verify-an-ai-answer): If the answer carries a proof, you can check it yourself — in your browser, offline, or from code. Here’s how.
- [AI hallucination: why confident AI is still often wrong](https://throndar.ai/learn/ai-hallucination-explained): Language models predict plausible text, not verified facts — which is why confidence and correctness are not the same thing.
- [Harvest now, decrypt later: the quantum threat, explained honestly](https://throndar.ai/learn/harvest-now-decrypt-later): The urgency isn’t a quantum computer today — it’s that data recorded today can be decrypted the day one arrives.
- [Content provenance for AI: C2PA, watermarking & signatures](https://throndar.ai/learn/ai-content-provenance-c2pa-signatures): Watermarking, C2PA Content Credentials, and cryptographic signatures each answer “where did this come from?” — with different tradeoffs.
- [Verifiable AI governance: the four signed legs auditors actually accept](https://throndar.ai/learn/verifiable-ai-governance-evidence): Map the four NIST AI RMF functions to four signed artifacts anyone can re-check — and know exactly where the proof stops.

## Optional

- [Full text](https://throndar.ai/llms-full.txt): this map plus the public page copy, plan and credit tables and article summaries
- [Sitemap](https://throndar.ai/sitemap.xml)