Frequently Asked Questions

AtlasProof is the public AI verification experience for the AirTrustOS trust platform. It lets anyone inspect how an AI response, claim, file or agent action would be verified — with a trust score, model agreement, evidence sources, risk flags, expert-review actions and a proof-certificate preview.

AtlasProof Beta v1 is a sandbox demo. The public experience is a fully-featured frontend shell; the verification result is a static demo. No real upload, provider call, remote engine call, expert escalation, billing or proof issuance happens in this release.

Open the Verify page, paste text, pick a verification mode (General, Medical, Finance, Legal, Research or Media), choose an assurance and model profile, and press Verify. In this demo the Verify button opens a detailed static demo result page.

No. The public verification experience and every marketing page are open. The Free plan needs no credit card, and there is no billing in this demo.

The homepage has a hero with a verification input and mode chips, a Try Proofence Engine section, an Explore the Platform card grid, Core Capabilities, Trusted Sources recommendations, Enterprise Partners, AI Startups & Service Providers, AI Agents/Models & Workflows, Expert Review tiers, Verified Experts, the Proofence Trust Architecture, an Enterprise section, a six-step 'How AtlasProof Verification Works' flow, Use Cases and a final enterprise CTA.

Trust Score (a single 0–100 score), Source Validation (checking claims against recommended trusted sources), Model Agreement (comparing how multiple models vote) and Risk Detection (flagging misinformation and low-confidence claims).

Six steps: Ask a Question, AI Models Respond, Trust Score Calculated, Sources Checked, Expert Review (optional) and Proof Issued. In V46 the 'Proof Issued' step is a proof-certificate preview only — no real cryptographic issuance.

Text and File Upload (PRO). Verification modes include General, Medical (PRO), Finance (PRO), Legal (PRO), Research (PRO) and Media Verification (PRO). Assurance modes are Fast, Balanced and Strict, and model profiles are Balanced, Strict and Expert.

The demo result shows a 0–100 trust score plus trust components: model confidence, source support, model agreement, media integrity, policy score and an uncertainty penalty. A governance action scale maps the score to APPROVED (≥80), WARN (60–79), REVIEW (40–59) or BLOCK (<40).

Multiple AI models each vote on the claim with a weight and a per-model score; the result page shows a weighted score, the majority verdict and whether the models disagree. In the demo these are static example values.

Risk flags highlight issues such as model disagreement, low confidence or domain-specific misinformation (for example, medical misinformation) with a severity and an explanation.

The result page lists evidence sources, AI model responses and related web sources as static demo examples. AtlasProof does not fetch live web, scrape pages or claim real-time retrieval in V46.

The result page offers Approve, Escalate, Block and Request Expert actions plus a reviewer-notes field across expert categories (Technology, AI, Legal, Medical, Finance, Security, General). In V46 these are frontend-only and submit nothing to a backend.

No. The proof certificate is a static demo preview (with a demo id, SHA-256 label, version and issued date). AtlasProof issues no real cryptographic proof, hash, signature or Merkle proof in V46.

Per-domain recommendations — for medicine: MedlinePlus, NICE, NHS, CDC, WHO, FDA and PubMed — grouped into official guidance, regulators/standards and literature. They are informational only and do not constitute professional advice.

Free, Pro, Expert and Enterprise. Pricing is illustrative and non-binding — there is no checkout or billing in this demo. Enterprise adds policy sets, workflow controls, multi-tenant governance and a future private Proofence Engine.

Hospitals & clinics, law firms, public authorities, universities, regulated enterprises and enterprise teams, plus AI startups and service providers who want a shared trust layer for their model outputs.

Use the Contact page to reach hello@atlasproof.com or hello@proofence.com, or visit www.proofence.com. There is no CRM or backend form submission in this demo.

No. The public verification experience is frontend-only in V46. Text is not sent to a real model, File Upload PRO does not upload or parse files, and no media, OCR, audio, video or image processing occurs.

No real file upload, media processing, OCR, provider or model execution, remote Proofence Engine call, expert-escalation backend, email/CRM, billing/credits mutation, proof issuance, credential storage or database writes. It never exposes secrets or raw environment values.

The public shell is the buyer-facing surface of the same AirTrustOS platform whose enterprise control plane lives under /admin and /verification — including the Proofence Engine adapter, engine request contracts and the engine evidence bridge, all of which remain sandbox-safe.

Account Console (/account) is a static preview of your AtlasProof account — verifications, proofs, expert requests, credits, subscription and settings. In V48 every value is demo data; there is no real account lookup, mutation, billing or credit action.

No. The role dashboards (/dashboards) are a roadmap preview. Role-based dashboard routing is roadmap-only in V48 — there is no auth and no role enforcement.

You can open the Sign Up preview (/sign-up), but authentication is not active in this V48 preview — nothing is submitted or stored. Continue with the Free Preview to explore the Account Console.

No. The Expert, Startup and Enterprise dashboard previews are static roadmap illustrations — no real review action, integration or tenant provisioning is performed.

No. Selecting a role only opens its static preview page. There is no role-based authorization or role-based data access in V48.

No. The API Keys page (/account/api-keys) is a preview only — no API key is created and no secret is shown. Real API access is roadmap.

No. Credits and Subscriptions pages are static previews — there is no billing, payment, checkout or credit mutation in V48.

Future waves would add real authentication, role-based authorization and per-role dashboard backends behind server-only, opt-in configuration and human approval — none of which is part of V48.

The Site CMS (V50) is a frontend content model for the public AtlasProof site — menus, footer, homepage sections, a homepage slider, public pages and a content hub (blog, news and trusted sources) with SEO metadata and categories. It is driven by a static module (lib/site-cms.ts); in V50 there is no database and no publish action.

They live in the header More menu and the footer, and open the public content hub: /blog, /news and /sources (each with detail pages), plus a /resources hub. Every article, news item and source is static demo content in V50.

No. The admin CMS editors (/admin/cms and its blog/news/sources/pages/menus/footer/homepage/slider/seo/categories previews) are non-submitting previews. Save Draft, Preview, Publish and Schedule are disabled — preview only, no DB write in V50.

The homepage shows a slider of featured slides read from the Site CMS model, using the AtlasProof primary color. It is a static, client-side preview — no autoplay backend and no DB read.

The sources (MedlinePlus, FDA, NIST, CISA, SEC Investor.gov, PubMed, WHO, NHS, NICE, EUR-Lex and more) are informational references only and do not constitute professional advice. External links open the official site; AtlasProof performs no fetch or scraping.

Real publishing requires V51 database-backed CMS (Postgres/Supabase/Neon) and Vercel Blob or a headless CMS for media. V50 ships only the static frontend foundation — no DATABASE_URL, Blob token or external CMS credential is needed.