# AiVACO > AiVACO Ltd is a UK company that builds sovereign AI and data systems and advises regulated organisations on AI assurance. Its platform, AAP (the AiVACO Assurance Platform), is a deterministic control layer that sits between any AI model and a regulated process: it grounds every output in approved sources, applies encoded rules, assigns an A1 to D4 risk tier, and records the result in a tamper evident audit trail before release. The control layer itself contains no AI. Markets: pharmaceutical and life sciences (GxP), education, government and public sector, and other regulated industries. AiVACO describes how its work aligns with regulatory frameworks (EU AI Act, EU GMP Annex 11 and draft Annex 22, ICH Q9(R1), GAMP 5, 21 CFR Part 11, UK GDPR). It does not claim endorsement or approval by any regulator. Draft texts are described as drafts. Contact: hello@aivaco.ai. Address: 8 New Road, Leighton Buzzard, LU7 2LX, United Kingdom. ## Platform and approach - [AAP, the AiVACO Assurance Platform](https://aivaco.ai/platform): AAP is a deterministic control layer that grounds, validates, risk tiers and records each AI output before release. Model agnostic, on your infrastructure. - [Why sovereign AI](https://aivaco.ai/sovereign-ai): Sovereign AI means you decide where data lives, which models touch it, who sees outputs and how long evidence is kept. How AiVACO makes that control real. - [Trust and security](https://aivaco.ai/trust): AiVACO security by design: network, identity, data, AI specific threats and supply chain controls, plus six responsible AI principles. - [How we engage](https://aivaco.ai/engage): Five steps from first briefing to scale, including the 90 day assurance pilot that runs in shadow mode and ends with an evidence pack. ## Advisory and training - [Advisory](https://aivaco.ai/advisory): Independent AI audits, feasibility studies, governance and validation for regulated organisations, from practitioners in regulation and AI engineering. - [Training](https://aivaco.ai/training): Role based AI training for leaders, Quality and IT teams, reviewers and staff. Eight programmes that leave records you can file in your QMS. - [AI Avatars](https://aivaco.ai/ai-avatars): Interactive AI presenters that answer only from approved content. Every response is grounded, validated, risk tiered and recorded by AAP before it is spoken. ## Industries - [Industries overview](https://aivaco.ai/industries): Different regulators, the same question: what did the AI do, what did it rely on and who was accountable. See how AiVACO applies to each sector. - [Pharma and life sciences](https://aivaco.ai/pharma): How AAP assures AI in GxP: Annex 11 and draft Annex 22, ICH Q9(R1), GAMP 5 and data integrity, with worked examples from Quality, QC and regulatory affairs. - [Education](https://aivaco.ai/education): AI that protects learners, respects their data and answers only from approved content. Safeguarding is routed to designated staff and every answer is recorded. - [Government and public sector](https://aivaco.ai/government): AI that explains how it influenced a decision, keeps citizen data under the department's control and avoids dependence on any single vendor. - [Other regulated markets](https://aivaco.ai/regulated-markets): The principles that make AI defensible in pharmaceutical GMP apply wherever a wrong output carries regulatory, safety or financial consequences. ## Insights - [Whitepaper: Building an AI Strategy You Can Defend](https://aivaco.ai/docs/whitepaper): A practical guide for Quality, validation, IT and governance leaders. - [Why the control layer should contain no AI](https://aivaco.ai/insights/why-the-control-layer-should-contain-no-ai): A control that uses AI to check AI inherits the weaknesses it is meant to catch. Why AiVACO builds the assurance layer from deterministic software. - [Draft Annex 22 in plain English for Quality leaders](https://aivaco.ai/insights/draft-annex-22-in-plain-english): What the first EU GMP annex on AI says, which uses it covers, what it allows and what it asks of you. A plain reading for Quality and validation leaders. - [Shadow mode: building evidence before you rely on AI](https://aivaco.ai/insights/shadow-mode-building-evidence-before-you-rely-on-ai): Shadow mode runs AI alongside an existing process without letting it affect decisions. How it works, what to measure and how to avoid the common mistakes. - [Sovereign AI for schools and multi academy trusts](https://aivaco.ai/insights/sovereign-ai-for-schools-and-multi-academy-trusts): What control over data, models, evidence and decisions means for a school or trust, and what governors and DPOs should ask before adopting AI. - [Transparent AI in government: what good evidence looks like](https://aivaco.ai/insights/transparent-ai-in-government-what-good-evidence-looks-like): For a public body, transparent AI means showing what it did in a specific case. What a record that stands up to scrutiny contains, and what to ask suppliers. - [The inspector is acquiring AI too](https://aivaco.ai/insights/the-inspector-is-acquiring-ai-too): Regulators are adopting AI themselves. What that changes about inspection readiness: from finding the information to being able to explain it. - [Intended use: the one document every AI system needs](https://aivaco.ai/insights/intended-use-the-one-document-every-ai-system-needs): An intended use statement says exactly what an AI system is for and what it must not be used for. What to put in it, with a worked example. - [Model agnostic by design: why your assurance should survive a model change](https://aivaco.ai/insights/model-agnostic-by-design): AI models change often. How to design assurance so the evidence stays valid when the model underneath is swapped, upgraded or withdrawn. ## Company - [Team](https://aivaco.ai/team): Meet the founders and the AI Lab behind AAP: pharmaceutical quality and regulatory expertise, AI engineering and commercial experience in regulated markets. - [Contact](https://aivaco.ai/contact): Tell us about your organisation and what you are exploring. A member of the AiVACO team replies within one working day. - [Privacy policy](https://aivaco.ai/privacy): How AiVACO Ltd collects, uses and protects personal data from this website, and the rights you have under UK GDPR. ## Optional - [Full text of the key pages and articles](https://aivaco.ai/llms-full.txt) - [Interactive demonstrator](https://demo.aivaco.ai/): uses fictional data and simulated AI responses.