About
Built from the ground up since 2021.
AIRAS Cloud exists because AI policy rarely survives contact with operational reality. Five years of deliberate technical and governance engineering sit behind the control plane - the risk model, the control architecture and the evidence trail.
- Founded
- 2021
- Team
- 8 specialists
- Based in
- Ireland
- Legal owner
- AFRH Consulting Limited
- Company number
- CRO 798243
- Built in-house
- Risk, controls, evidence
Our position
Responsible AI adoption is not a choice between speed and control. Organisations move faster when the route to approval is clear, consistent and evidenced.
AIRAS Cloud is designed around the operational layer of governance: registering AI use, assessing risk under approved rules, routing the right reviewers, collecting evidence as work happens and retaining a decision trail that stands up to scrutiny.
The product is shaped by experience in regulated manufacturing, financial services, healthcare and critical national infrastructure, where quality, privacy, security and audit obligations sit alongside commercial pressure to adopt new capability.
Built from the ground up since 2021 - architecture, risk logic, control libraries and evidence model designed in-house under governed change control. Five years of development and a full alpha programme with regulated Irish organisations have kept the product grounded in real governance obligations rather than abstract maturity models. AIRAS Cloud is an internally verified platform. Independent assurance is not yet complete, and its status is published openly in the trust centre.
Since 2021
A deliberate build, year by year.
Nothing here was assembled quickly. Each layer was designed, tested against real governance obligations and version-controlled before the next was added.
2021
The problem, framed properly
AIRAS began as a working response to a recurring failure in regulated programmes: AI policy existed on paper, but nothing carried an assessment through to a defensible, evidenced decision. The first risk model and control taxonomy were drafted and tested against real governance obligations.
2022
Deterministic risk engine
The versioned scoring model, mandatory floors, prohibited-use checks and explanation trace were designed and iterated. The founding principle was set early: AI does not approve AI.
2023
Evidence and lifecycle model
Control libraries, evidence requirements, reviewer routing with segregation of duties, and append-only audit history were built out into a single connected record structure.
2024
Platform engineering and governed change
The control plane was engineered as a multi-tenant product with configuration governance, ruleset versioning, integration surfaces and a working synthetic demonstration environment.
2025
Agents, runtime and monitoring
Coverage extended beyond registration and approval into agent oversight, runtime moderation, incident handling, periodic reassessment and value realisation.
2026
Consolidated and internally verified
AIRAS Cloud is a tested control plane under governed change control. The full lifecycle — intake, deterministic risk assessment, controls, review, decision, monitoring and evidence export — is implemented and exercised by automated regression, with alpha feedback from regulated organisations in Ireland.
Where ARIE began · 2022 to 2024
Built by hand, from the Act itself.
Before there was a platform, there was a spreadsheet. We read printed copies of the EU AI Act and built our own risk categorisation calculator, one clause and one formula at a time. No AI wrote it. These are a few excerpts from that original workbook.
If any = Yes → Category = PROHIBITED
What we were thinking: the law is not a score. If a practice is banned, no number should be able to argue it back in. That rule is still the first gate in ARIE.
Bands: 0–24 Minimal · 25–59 Enhanced · ≥60 Enhanced, Critical
What we were thinking: the less a person is involved, the more risk you carry. Points only apply once the legal gates are clear, and every factor came with a notes column for the evidence behind it.
Ten principles, each with its own weight
Every factor was tagged to the principles it touches, then weighted. Safety counted double. That reviewer view was locked so only the governance committee could see it.
- P1Safety2.0×
- P7Data protection and privacy1.8×
- P9Competition law1.7×
- P4Avoid bias and discrimination1.6×
- P6Secure and resilient1.6×
- P2Adequate human oversight1.5×
- P3Accuracy and reliability1.5×
- P5Transparency and explainability1.2×
- P8Intellectual property1.1×
- P10Sustainability1.1×
2022
The Act, on paper
Printed copies of the draft EU AI Act, read clause by clause. Every prohibited practice and Annex III trigger was written down as a plain yes or no question.
2023
The master workbook
Eighteen tabs, worked left to right: intake, prohibited check, Annex III, scoring factors, decision. Law gates first; points only count when no legal category applies.
2024
Principles and committees
Ten weighted principles, a locked reviewer sheet for the governance committee, and category guidance telling teams exactly what each outcome obliges them to do.
Today
ARIE
The same thinking, rebuilt as a deterministic, versioned engine with evidence, segregated review and an audit trail. Still no AI deciding the outcome.
Excerpts are reproduced from the founders' original workbook for illustration. The workbook itself is not published.
Leadership
Founders who own the detail.
Richard Higgins
Founder and Chief Executive
Creator of AIRAS and a successful entrepreneur with more than fifteen years building e-commerce platforms and regulated tooling. Richard has led complex technology, governance and transformation programmes across regulated and enterprise environments, and has built AIRAS from the ground up since 2021 - the risk model, the control architecture and the evidence design.
Connect on LinkedIn to read public recommendations and testimonials.- Product architecture
- Risk and control design
- Enterprise delivery
Amy Feery
Co-Founder and Chief Product Officer
Amy brings fifteen years of product management and product ownership, specialising in AI governance. She owns the product operating model, the governance lifecycle experience and how regulated teams actually work inside AIRAS day to day.
- AI governance practice
- Product ownership
- Customer operating model
A team of 8
Deliberately small and deliberately senior, spanning engineering, data and risk, product and governance, and commercial delivery. Ireland, working with organisations across Europe.
See open rolesEngineering
Platform, control plane, integrations and secure delivery.
Data and risk
Risk modelling, ruleset versioning, analytics and evidence integrity.
Product and governance
Governance practice, lifecycle design and configuration standards.
Commercial and delivery
Design partners, enterprise engagement and programme delivery.
Experience developed across complex enterprise environments
AIRAS Cloud is founded on experience delivering complex technology, governance and transformation programmes across regulated and enterprise environments.
Aer LingusAviation
AvolonAviation leasing
AIBBanking
Bank of IrelandBanking
IBMTechnology services
KyndrylTechnology services
EirGridCritical infrastructure
Virgin MediaTelecoms
Irish GovernmentPublic sector
Bring AIRAS Cloud into your organisation
A control plane for AI governance, built out from a 2021 concept, internally verified and shaped by an alpha programme with regulated Irish organisations.
No commercial commitment. No confidential information required.