VERA 1.0: AI answer verification for confidential professional work

VERA (Verified Evidence & Reasoning Architecture) is a multi-model AI verification platform for professionals bound by confidentiality: lawyers, legal advisers, notaries, doctors, healthcare professionals, board members and estate agents.

One AI model can be convincingly wrong. VERA lets you check the facts, challenge the reasoning and review the sources before you act on an AI-generated answer.

🌐 Website: https://iamvera.ai 🚀 Start your environment: https://saas.iamvera.app (first month free, 10 runs, no credit card required)

---

✦ VERA SaaS platform: multi-model AI verification

VERA routes each question through a configurable verification chain in which every model has a distinct role:

Step	Model	Role
1	Claude	Drafts the initial answer
2	GPT	Reviews factual claims for errors and missing support
3	Grok	Challenges assumptions and reasoning
4	Perplexity	Verifies claims against current public sources

• Radical transparency: every correction, disagreement and source behind the final answer stays visible.
• No false consensus: when models disagree, VERA surfaces it instead of hiding it.
• Configurable per session: choose which models participate and see tokens and cost per model call.
• Pay per verified question: one question through the chain is one run, not billed per token.
• Vera Office: privacy-preserving AI document analysis and validated, controlled editing in a browser-based DOCX workspace.
• Published evidence: test runs of the verification chain, including failures and raw JSON data, at https://iamvera.ai/evidence/

🛡️ Semantic Privacy Shield: pseudonymisation before external AI processing

The Semantic Privacy Shield protects confidential documents inside the customer-isolated VERA environment:

1. Detect and pseudonymise: names, dates, case numbers, diagnoses and other sensitive values are replaced with synthetic values, while roles and relationships stay intact so the AI still understands the file.
2. Fail-closed outbound check: before anything is sent, the prepared text is checked. If a detected real value remains, the request is blocked. No fallback, no silent override.
3. Restore locally: the original values are restored inside the protected environment after the answer returns.

More: https://iamvera.ai/privacy-shield/

🔍 Vera X-Ray: on-device detection of sensitive data (NEW)

Vera X-Ray adds a privacy boundary before document context reaches VERA, directly in your browser.

• Detect locally: PDFs, scanned PDFs (local OCR), TXT and Markdown, in Dutch and English.
• Deterministic rules, always on: IBAN, BSN, phone numbers, email, postcodes, licence plates, card numbers, dates, amounts, case and land-registry numbers, passwords and keys.
• Optional local recognition model: a one-click install, running in the browser via WebGPU or WASM, for names, organisations, places, health information and special categories of personal data. Nothing is downloaded without your choice; model files are verified against pinned SHA-256 digests.
• Human review before release: inspect findings, mark what was missed and approve what goes out.
• Reversible tokens: random tokens (not hashes); mappings are stored in an authenticated encrypted envelope and can be PIN-protected.
• Restore on-device: VERA answers with references; real values are put back on your device only.

📦 Chrome Web Store: https://chromewebstore.google.com/detail/vera-x-ray/aelkgfaihpkdibennlolmafndbjginip A Firefox version is in preparation.

More: https://iamvera.ai/x-ray/

---

📚 Insights: VERA Blog & Newsletter

VERA Blog: daily analysis on AI verification, AI governance, the EU AI Act and compliance, privacy and GDPR, AI security and agentic AI, written for lawyers, notaries, compliance teams and public-sector professionals who act on AI output. Available in English and Dutch. → https://iamvera.ai/blog/ · https://iamvera.ai/nl/blog/

The IamVera Newsletter: a periodic review of AI assurance, multi-model verification, governance and privacy-aware document workflows, bundling the most relevant analyses into one edition. → Read past editions: https://iamvera.ai/newsletter/ → Subscribe on LinkedIn via the newsletter page

---

🇪🇺 Built in the EU, independently certified

• Built, operated and hosted in the EU
• Certified to ISO/IEC 27001, ISO 9001, ISO/IEC 42001 and NEN 7510
• Transparent model routing: approved, transformed context is sent to the selected AI providers via OpenRouter; those providers may process data outside the EU

⚠️ Scope and limitations

VERA is a verification layer, not a guarantee of truth. It does not replace professional judgement, and final responsibility for acting on an answer remains with the user. The Privacy Shield and X-Ray reduce exposure but do not claim perfect detection or full anonymisation: indirectly identifying combinations of facts can remain, and OCR-derived text needs extra review.

---

Links: Website · Pricing · Evidence · Blog · Newsletter · Contact

© 2026 The Coding Company B.V.