Home / Software Audit (SAM)
Software Asset Management · AI compliance
Know your licence position before the vendor does.
Independent software audits that compare what you are entitled to with what you actually run, then reduce the gap and the risk. Now including how your organisation uses AI.
What we offer
Software audit services
Licence compliance audit
A full Effective Licence Position (ELP) per vendor: entitlements against deployments, with gaps and surpluses.
Vendor audit support
Preparation and support during audits from Microsoft, Oracle, IBM, Broadcom (VMware), SAP, Adobe and others.
Licence optimisation
Remove shelf-ware, reharvest unused licences and choose the right metric, edition and subscription tier.
Inventory & discovery
Hardware and software inventory across servers, virtual platforms, cloud and end-user devices.
Cloud & SaaS review
Microsoft 365, Azure and SaaS subscriptions checked against real usage and assigned users.
SAM programme
Processes, roles and tools aligned with ISO/IEC 19770-1, so compliance stays current after the audit.
Audit method
Four steps to an Effective Licence Position
Scope
Vendors, legal entities, environments and the contracts in force.
Discover
Inventory data collected with scripts or SAM tools, plus virtualisation and cluster topology.
Reconcile
Deployments matched to entitlements under each vendor’s licence metrics and product use rights.
Report
ELP report, financial exposure, savings and a prioritised remediation plan.
AI audit & compliance
Audit how your organisation uses AI
AI now arrives inside the software you already license, as SaaS assistants your staff sign up for, and as models your teams build. We extend the software audit to cover it: what AI is in use, under which terms, how it is secured and which rules apply.
AI inventory & shadow AI
Discovery of AI assistants, AI features in licensed products, AI SaaS subscriptions and in-house models, with owners and data flows.
EU AI Act readiness
Classification of each AI use case by risk level, your role as provider or deployer, and the obligations that follow: AI literacy, transparency, documentation and human oversight.
ISO/IEC 42001 gap assessment
Gap analysis against the AI management system standard, aligned with your existing ISO/IEC 27001 controls and NIS2 obligations.
AI security review (ISO/IEC 27090)
Review of development-time and runtime threats to your AI systems, following the guidance in ISO/IEC 27090, with mitigations mapped to your controls.
AI licences & usage terms
AI add-on seats and consumption-based billing checked against real usage, plus the data processing and IP terms of each AI provider.
GDPR & data exposure
What personal and confidential data reaches AI tools, where it is processed and whether a DPIA is needed.
AI security review scope
Threats we assess, by lifecycle stage
| Stage | Threats | What is at risk |
|---|---|---|
| Development time | Training data leak, data poisoning (direct or via supply chain), model leak, model poisoning | Training data confidentiality, model behaviour, intellectual property |
| Runtime: input | Prompt injection, evasion, model exfiltration, model inversion, membership inference, disclosure in output, AI resource exhaustion | Model behaviour, training and input data, IP, availability |
| Runtime: data & model | Augmentation data leak or manipulation (RAG), runtime model leak or poisoning, input data leak | Input and augmentation data confidentiality, model behaviour |
| Runtime: conventional | Output containing injection payloads, SQL injection, credential attacks on the AI application and its infrastructure | Application, infrastructure and downstream systems |
Sample deliverable
What an ELP summary looks like
| Product | Metric | Entitled | Deployed | Position |
|---|---|---|---|---|
| Windows Server Datacenter | 2-core packs | 48 | 64 | −16 gap |
| SQL Server Enterprise | 2-core packs | 12 | 8 | +4 surplus |
| Oracle Database EE | Processor | 4 | 4 | compliant |
| Microsoft 365 E3 | User | 450 | 382 | +68 surplus |
- A vendor audit letter has arrived, or a renewal is coming up
- Virtualisation or cloud migration changed how licences are counted
- Mergers, new entities or headcount changes
- Budget pressure and suspected over-licensing