AI Governance Consulting
AI governance consulting that puts the controls inside your AI
We inventory every AI system you run, classify its risk under the EU AI Act, and build the evidence auditors and customers ask for: policies, evaluation sets, monitoring and audit logs. One control set covers ISO/IEC 42001, NIST AI RMF and the EU AI Act.
- 3-in-1 frameworks covered by one control set
- 2027 EU AI Act high-risk rules apply from 2 December
- 2028 AI in medical devices covered from 2 August
- 5.0 Clutch rating, 22 client reviews
Who calls us
Governance help for four owners of AI risk, from code to boardroom
- 01
Head of AI or CTO
Models are live, but nobody can list them all. We build the inventory and put monitoring where it is missing.
Map our AI estate - 02
Risk or legal lead
You own the policy. We turn it into controls that engineers can implement and auditors can test.
Turn policy into controls - 03
Medtech or pharma QA
AI inside a device or a GxP process. We align AI controls with IEC 62304, ISO 14971 and your QMS.
Talk regulated AI - 04
AI vendor in the EU
Buyers send AI questionnaires. We prepare the documents and evidence that answer them once.
Prepare for buyer audits
Overview
AI governance consulting in four answers
Who needs it
Companies running or buying AI that must show regulators, auditors or customers how each system is controlled.
What changes
Every AI system gets an owner, a risk tier, tested controls and evidence you can hand to an auditor.
Where we start
A free 30-minute triage call. We list your AI systems with you and flag which ones carry regulatory exposure.
How it is priced
Gap assessment from €5,000£4,700$6,000, a fixed fee for 3–4 weeks. Control building and ongoing assurance are quoted from its findings.
Is this for you
Signs you need AI governance now and signs you can wait
Start now if
- AI already makes or shapes decisions about customers, patients or employees.
- Enterprise buyers or notified bodies ask how your AI is tested and monitored.
- You are preparing for an ISO/IEC 42001 certification audit.
- Nobody can say how many AI tools and models your teams use today.
Not the right service if
- You have no AI in use yet. AI Consulting is the better first step.
- You need a legal opinion. We work alongside your counsel, not instead of it.
- You want us to certify you. Certificates come from accredited bodies; we get you audit-ready.
- You only need a governance software tool. We set up controls in your systems and don’t resell platforms.
How we work
How does an AI governance engagement run? Five steps from inventory to audit
Inventory, Classify, Gap-check, Control, Operate. The first three make up the fixed-fee assessment. Controls are then built into your systems and kept current as models, vendors and regulations change.
- 01
Inventory
We list every model, AI feature and third-party AI tool in use, including the ones IT never approved.
Output: an AI system register with owners and vendors.
- 02
Classify
We assign each system an EU AI Act risk tier and note which frameworks and sector rules apply.
Output: a risk classification with the reasoning written down.
- 03
Gap-check
We compare current practice with ISO/IEC 42001, NIST AI RMF and the AI Act duties that apply.
Output: a gap report ranked by regulatory exposure.
- 04
Control
We write the policies and build the technical controls: evaluation sets, monitoring, logging, human review.
Output: a control matrix linked to evidence.
- 05
Operate
We review new AI systems, incidents and rule changes, and refresh the evidence.
Output: an audit-ready evidence pack, kept current.
What you receive
What will auditors see? Evidence, not slideware
Each document is tied to a running control, so it stays true after the engagement ends. Formats match what your QMS, ISMS or GRC tool already uses.
- AI system register
- Every model and AI tool, its owner and its vendor.
- Risk classification
- EU AI Act tier and the sector rules that apply.
- Control matrix
- One control mapped to several frameworks.
- AI policy set
- Acceptable use, procurement, incident response.
- Evidence pack
- Test results, logs and model documentation.
Building new AI too? AI Consulting and AI Transformation include these controls from day one.
Scope of work
What we cover
Pick the parts you need. Most clients start with inventory and classification, then add controls.
- Shadow AI discovery Unapproved tools found through interviews, SSO logs and spend data.
- EU AI Act readiness Risk tiers, provider vs deployer duties, deadlines.
- ISO/IEC 42001 prep An AI management system ready for the certification audit.
- Model validation Bias, robustness and drift tests on your data.
- GenAI and agents Guardrails, red-teaming and action limits for LLM tools.
Engagement options
How is AI governance consulting priced?
The triage call is free. The gap assessment is a fixed fee from €5,000£4,700$6,000 for 3–4 weeks, agreed before you sign. Control building and ongoing assurance are quoted from what the assessment finds.
Triage call
€0£0$0
30 minutes to list your AI systems and spot the ones with regulatory exposure.
No preparation needed
Gap assessment
From €5,000£4,700$6,000, fixed fee
Inventory, risk classification and gap report against ISO/IEC 42001, NIST AI RMF and the AI Act.
3–4 weeks · documents you keep
Ongoing assurance
Quoted after the assessment
Quarterly reviews, new-system onboarding and evidence kept audit-ready.
Priced from your system register
Every assessment includes: a readout for your board or risk committee, the full system register and a prioritized fix list.
Control building: priced per control set once the gaps are known, so you pay only for what your risk profile requires.
What sets the fee: the number of AI systems and business units in scope. We fix it in writing after the triage call.
Confidentiality: we sign an NDA before the assessment, and a data processing agreement, or a HIPAA business associate agreement, before we handle personal data or PHI such as single sign-on logs.
Medical devices or GxP? We align AI controls with your QMS, IEC 62304 and ISO 14971, so one audit trail covers both the device rules and the AI Act. For GxP processes we work to GAMP 5 (second edition), which covers AI and machine learning, and to your validation procedures.
Prices exclude VAT where it applies. We are not a law firm or a certification body; legal opinions and certificates come from your counsel and accredited auditors.
AI Governance or AI Consulting? Pick by where your AI is today.
| Compare | AI Governance | AI Consulting |
|---|---|---|
| Starting point | AI already in use or about to launch | No AI yet, or unsure where it pays off |
| Main question | Can we prove our AI is under control? | Where should we use AI first? |
| Main output | Risk register, controls and audit evidence | A roadmap for the first AI build |
| Pricing | Gap assessment from €5,000£4,700$6,000, fixed fee | Discovery sprint from €5,000£4,700$6,000, fixed fee |
Controls in production
Oversight we have already built into live AI
Governance is easier when the people writing controls have shipped AI themselves. These systems run with human review, validation and monitoring we designed. Each figure is quoted from a published case study.
- 83% of raw claims converted to a rebate-eligible format with no manual stepHealthcare analytics firm, USA
Human-in-the-loop design for drug-code matching on rebate claims
The other 17% are flagged automatically for human review. See the case: rebate-claims engine for a US healthcare data company
- 1,000+ staff hours freed every yearNHCA · post-acute care, six US states
Traceable financial reporting from one governed data warehouse
Reports ready 50% sooner; invoice and payment delays down 45%. See the case: NHCA reporting automation
- 3× quicker catalog onboardingE-commerce · corporate gifts
AI-written product copy with duplicate and quality checks
Manual catalog effort cut by over 45%. See the case: corporate-gifting catalog pipeline
- 99% detection accuracy for rust vs. non-rust surfacesIndustrial inspection, USA
Corrosion detection validated under real-world conditions before it feeds a digital twin
Manual inspection spend down 60%. See the case: drone-imagery corrosion detection
- ±1.5 mm tolerance held, cycle time unchangedGentherm · tier-1 automotive
Vision inspection revalidated after the line and lighting changed
False alarms reduced and validated on the running line. See the case: Gentherm computer-vision inspection
How we compare
Law firm, Big-4, software platform or engineers? Who makes governance actually run?
A law firm interprets the rules, an audit firm tests against them, and a platform tracks tasks. Someone still has to build the controls inside your AI systems. That is the part we do.
| Factor | Blackthorn AI governance | Law firm | Big-4 or audit firm | Governance platform |
|---|---|---|---|---|
| Main output | Register, control matrix and evidence tied to running systems | Legal analysis and policy drafts | Assessment report and audit findings | Workflows and dashboards to fill in |
| Technical controls | Built by our engineers: evaluation, monitoring, logging, review | Out of scope | Recommended, rarely built | Tracked, built by your team |
| Team | ML engineers and scientists who ship AI | Lawyers and policy specialists | Risk consultants and auditors | Your own staff, with vendor support |
| Medtech and GxP | IEC 62304, ISO 14971 and GxP experience in-house | Depends on the firm | Usually a separate practice | Generic templates |
| Afterwards | Controls keep running in your systems | Advice on request | Next audit cycle | Licence renewal |
Comparison reflects typical engagement scopes, not specific firms.
Our difference
Governance written by people who ship AI
- Engineering first. Every policy line maps to a control we can show running in code or infrastructure.
- Regulated pedigree. We already ship software under IEC 62304, ISO 14971 and GxP rules.
- One control, many frameworks. A single control set answers the AI Act, ISO/IEC 42001 and NIST AI RMF.
- Tool-neutral. We work in your GRC, QMS or ticketing tools; no new platform required.
- Honest limits. We prepare you for audits and legal review; certificates and legal opinions stay with accredited bodies and counsel.
Who runs your governance work?
A lead ML engineer owns the technical controls, a scientist designs validation and bias testing, and a regulatory specialist maps everything to the frameworks you answer to. For devices and GxP, our QA team joins.
- Alex Gurbych, PhD, Co-founder and Director
- Serhiy Protsenko, DrSci, Prof., Principal Data Scientist
- Dmytro Krasnienkov, PhD, Principal Scientist, Bioinformatics
- Ivan Izonin, PhD, DrSci, Scientific Advisor, Artificial Intelligence
Clutch feedback
How clients describe working with us
22 verified reviews, rated 5.0
- Verified on Clutch
«Blackthorn.ai is great at adapting to our needs»
It’s not easy to work with us as a critical infrastructure provider because we have many restrictions that make it difficult to work with data since it has to be secured.
Under NDA, Deputy, Surgical Center
- Verified on Clutch
«The team surpassed expectations on timelines»
…provided much needed guidance and overall input on design, all while operating with a high degree of autonomy.
Carl Kaub, Vice President of Chemistry, HTG Molecular Diagnostics
- Verified on Clutch
«What impressed us the most»
…was their ability to take over a fragile frontend in production, quickly fix a critical authentication issue, and then deliver a large UI redesign with strong QA discipline, all while keeping scope controlled and avoiding risky backend changes.
Egoitz Arruti, CTO, Ubikare
01 What is AI governance?
AI governance is the set of owners, policies and technical controls that decide how AI systems are approved, tested, monitored and retired. Good governance produces evidence: anyone can check which AI you run, how risky it is and how it is controlled.
02 What do AI governance consulting services include?
An inventory of every AI system you run, a risk classification under the EU AI Act, a gap report against ISO/IEC 42001 and NIST AI RMF, and then the controls themselves: policies, evaluation sets, monitoring, logging and human review. At Blackthorn AI the first three make up a fixed-fee assessment of 3–4 weeks; control building and ongoing assurance are quoted from what it finds.
03 What is an AI governance framework?
A structured set of roles, policies, risk methods and controls for building, buying and running AI. The three most used are ISO/IEC 42001, an AI management system standard you can be certified against; the NIST AI Risk Management Framework, a voluntary US framework; and the EU AI Act, which is law. We build one control set and map it to each framework you answer to. Banks and insurers also answer to model risk rules, such as SR 11-7 in the US or the PRA’s SS1/23 in the UK, and we map the same controls to them.
04 Does the EU AI Act apply to my company?
It applies if you place AI systems or general-purpose AI models on the EU market, use AI systems in the EU, or sit outside the EU while the output of your AI is used there. Your duties depend on your role, provider or deployer, and on each system’s risk tier: prohibited, high-risk, transparency or minimal. Fines reach €35 million or 7% of worldwide annual turnover for prohibited practices. The triage call gives a first read on which of your systems carry duties; a legal opinion comes from your counsel.
05 What is the difference between a provider and a deployer under the EU AI Act?
A provider develops an AI system, or has it developed, and places it on the market or puts it into service under its own name. A deployer uses an AI system under its own authority at work. For high-risk systems, providers carry most of the duties: risk management, data governance, technical documentation, logging, human oversight design, accuracy and robustness testing, a quality management system and a conformity assessment before launch. Deployers must use the system as instructed, assign trained people to oversee it, monitor it, keep its logs for at least six months and tell workers before using it at work. A company that puts its name on a high-risk system or changes it substantially becomes its provider.
06 When do the EU AI Act high-risk rules start to apply?
The AI Act (Regulation (EU) 2024/1689) entered into force on 1 August 2024. Bans on prohibited practices apply from 2 February 2025 and rules for general-purpose AI models from 2 August 2025. The Digital Omnibus on AI (Regulation (EU) 2026/1744) moved the high-risk rules: they apply from 2 December 2027 to stand-alone systems listed in Annex III, such as AI used in hiring, credit scoring or education, and from 2 August 2028 to AI built into products covered by Annex I, such as medical devices. The duties themselves did not change, so the time until then is for building controls.
07 What is the difference between NIST AI RMF and ISO/IEC 42001?
NIST AI RMF 1.0, published in January 2023, is a voluntary US framework that organizes AI risk management into four functions: Govern, Map, Measure and Manage. There is no certificate for it. ISO/IEC 42001:2023 is an international standard for an AI management system, with requirements that an accredited body audits and certifies. Many controls overlap, so we build each one once and map it to both, plus the EU AI Act duties that apply to you. ISO/IEC 42001 uses the same management-system structure as ISO/IEC 27001, so it can share your ISMS policies, internal audits and management reviews.
08 Can you certify us for ISO/IEC 42001?
No. Certificates are issued by accredited certification bodies, and the firm that builds your system should not certify it. We get you audit-ready: the AI management system, policies, risk and impact assessments, and evidence that the controls run. You then choose a certification body for the audit.
09 How long does an AI governance assessment take?
3–4 weeks: the inventory in week 1, the risk classification in week 2 and the gap check in weeks 3–4. It ends with a gap report ranked by regulatory exposure and a readout for your board or risk committee. Building controls takes longer and depends on how many gaps the assessment finds. From your side it needs a named contact, short interviews with each team and read access to single sign-on, spend and SaaS admin reports.
10 How much does AI governance consulting cost?
The 30-minute triage call is free. The gap assessment is a fixed fee agreed before you sign: from €5,000 in Europe, £4,700 in the UK and $6,000 in other regions, excluding VAT where it applies. Control building is priced per control set once the gaps are known, and ongoing assurance is quoted from the size of your system register.
11 How do you govern generative AI and AI agents?
With the same register and risk tiers as other AI, plus controls for the ways LLMs fail: evaluation sets for accuracy and harmful output, red-teaming for prompt injection and data leaks, guardrails on inputs and outputs, limits on what an agent may do on its own, human approval for high-impact steps, and logs of every prompt, tool call and result. Our engineers also build AI agents, so we test these controls in the code.
12 What is shadow AI and how do you find it?
Shadow AI is AI used without approval: chatbots, browser plug-ins, AI features switched on inside SaaS tools, or models a team deployed on its own. We find it through short interviews with each team, single sign-on (SSO) logs and software spend data, then add each tool to the register with an owner and a decision: approve, restrict or replace. Our guide to shadow AI covers the six places it hides and how to detect each one.
13 How does AI governance fit with IEC 62304 and ISO 14971?
For AI in a medical device, IEC 62304 governs the software life cycle and ISO 14971 governs risk management. We treat the model as part of the device software: model risks such as drift or bias go into your ISO 14971 risk file, model changes follow your IEC 62304 change control, and validation records sit in your QMS. One audit trail then serves your notified body and the AI Act duties that apply to devices from 2 August 2028. In the US, planned model updates can be set out in advance in a predetermined change control plan (PCCP), as the FDA’s final guidance of December 2024 on AI-enabled device software describes.
14 How does AI governance fit with GDPR?
When an AI system uses personal data, GDPR applies alongside the AI Act. High-risk processing needs a data protection impact assessment (GDPR Article 35), and people have rights when a decision with legal or similarly significant effects is made solely by automated means (Article 22). Some deployers of high-risk AI, such as public bodies and companies that score credit or price life and health insurance, must also run a fundamental rights impact assessment under the AI Act (Article 27). We keep one AI system register and evidence pack that your DPO can reuse for each of these.
15 Do you replace our legal counsel?
No. Lawyers interpret the law and give legal opinions. We are engineers and scientists who turn that interpretation into controls that run in your systems, and into the evidence that proves it. We work alongside your counsel, and often your auditors, on the same register and control matrix.
16 Can law firms, consultancies or MSPs work with you?
Yes. Law firms and consultancies bring us in to build and test the technical controls their advice calls for, and MSPs and integrators can include our assessment or control building under their own contract. Book a call to agree the scope, the price and how we appear to your client.
List your AI systems with us in 30 minutes
Bring what you know about the AI your teams use. We map it on the call and send a short note on where your regulatory exposure sits.