We help you understand what AI is already doing inside your organization, what it can access, where the real exposure is, and how to safely expand the use of AI.

A practical, evidence-backed review of how AI is being used, what it can access, where the material risks are, and how to expand adoption safely.
It is aligned to NIST AI RMF / the NIST Generative AI Profile and informed by ISO/IEC 42001 management-system principles. It is not an ISO/IEC 42001 certification audit, not a generic compliance checklist, and not solely an AI penetration test.
The AI inventory and risk ratings feed EU AI Act risk classification. They are not a statement of compliance with the Act.

Most organizations are deploying AI faster than they're governing it. The gap between adoption and oversight is where regulatory, reputational, and operational risk lives.
Marcus Atkins, Director of Compliance Services
Establish business objectives, stakeholders, important use cases, sensitive data, regulatory/contractual constraints, systems in scope and the evidence plan.
Establish business objectives, stakeholders, important use cases, sensitive data, regulatory/contractual constraints, systems in scope and the evidence plan.
Nine control domains connect technical exposure to executive accountability.
For AI you build or connect yourself, we also check who can retrieve what, how secrets are handled, defenses against prompt injection, how outputs are validated, what agents and tools are allowed to do, secure development and whether an agent can be switched off quickly.
Framework alignment is a crosswalk, not a certification claim.
We verify the current admin settings for the platforms you actually use.
ChatGPT
SSO/SCIM • RBAC • apps/connectors • GPT sharing • web/agent/developer features • retention • compliance logs
Gemini
OU/group scope • history/retention • Workspace data access • Gem/conversation sharing • OAuth • audit/usage
Claude
SSO/domain/SCIM • retention • projects • integrations/MCP • code access • tenant/network controls • audit/Compliance API
Microsoft 365 Copilot
Entra controls • permission hygiene • Purview labels/DLP • agent/connector governance • audit/eDiscovery • lifecycle
Vendor settings are verified against current documentation during each engagement.
The assessment ends in decisions, not a 100-page audit report.
Turns technical/governance evidence into an executive narrative: footprint, top findings, heatmap, control posture, platform benchmarks and prioritized roadmap.
Turns technical/governance evidence into an executive narrative: footprint, top findings, heatmap, control posture, platform benchmarks and prioritized roadmap.
For qualified prospects. Contact us to learn more.
A no-obligation first look using read-only evidence. Within a week you get a short readout of what we found, a preliminary heatmap, and a recommendation on whether a full assessment is warranted.
We would rather show you than tell you.
Effort for you: about 3–4 hours (sponsor 1.5 h, admin 1.5–2.5 h).
* Offered to a limited number of qualified organizations each quarter, for a nominal fee of $1 under a formal engagement letter and NDA.
3–4 Weeks
Establish the footprint, validate the exposure, prioritize the fixes and give the business a safe path forward
Senior-Led
The people in the room read the logs, validate the tenant and brief your executives
Fixed-Price
Clear scope boundaries that protect your budget
Evidence-Driven
Questionnaires are one input, not the answer
Most organizations have more AI usage, integrations and autonomy than their existing governance can see.
Shadow usage
Employees adopt consumer and niche AI tools outside procurement, identity and logging.
Data reach
Approved AI can inherit access to email, documents, code, CRM and other sensitive systems.
Agentic actions
Connectors and agents can move from “answering” to sending, creating, changing or deleting.
Configuration drift
Enterprise licenses do not automatically mean least privilege, safe sharing, good retention or monitoring.
Governance gap
Policy, legal, security and business decisions are often disconnected or slow.
Opportunity cost
Blanket bans and slow approvals push users to workarounds and suppress legitimate productivity gains.
Enable, do not just block
Every risk comes with a safer path to the business outcome.
Discovery triangulates multiple signals
We do not rely on users remembering, or admitting, every AI tool they use.
Configuration, not just policy
An enterprise license is not a secure deployment; we check the actual settings.
Framework-ready
Aligned to NIST AI RMF and informed by ISO/IEC 42001, so later programs build on this work.

The problem is not AI itself. It is AI that nobody manages, with unclear data boundaries and inconsistent controls.
The AI Exposure Assessment is the technical, security-focused check of the AI tools actually in use. AI Governance & Assurance and ISO 42001 are about governance and certification.
"Do we have a problem?"
For qualified prospects. Contact us to learn more.
"What AI is actually running, what can it reach, what do we fix first?"
"How do we run AI governance day to day?"
"How do we prove it to customers?"
* Offered to a limited number of qualified organizations each quarter, for a nominal fee of $1 under a formal engagement letter and NDA.
We've sat in the CISO seat. Our recommendations come from hands-on experience building security programs, not theoretical frameworks.
Exposure findings, AI governance and ISO 42001 readiness sit with one team, so the work carries forward without a handoff.
SOC 2, ISO 27001, PCI DSS, NIS2, DORA, GDPR, HIPAA, CMMC, ISO 42001: one team, full coverage, no handoffs.
Next step: a 30-minute scoping call. With your executive sponsor and security/IT owner, we confirm your likely AI footprint, the use cases that matter most and the evidence available, and leave you with a fixed-scope proposal. Useful even if you decide to do the work yourselves.