Dallas, Texas · Technology consulting since 201635 services · 3 practices · 8 industries Technology in ActionPartnerships817-968-0523
Practice 02 · AI, cloud & digital trust

Trust you can demonstrate.

Customers ask for SOC 2, ISO certificates, FedRAMP and now a straight answer on AI. We build the governance behind those answers, assess it and test the technology.

  • ISO/IEC 42001
  • SOC 1 · SOC 2
  • FedRAMP · CMMC
IDENTITYDATAMODEL Human oversight Risk classification Model evaluation
Service lines
12 across AI governance and cloud assurance
Typical clients
SaaS, AI and cloud providers; regulated buyers
Frameworks
ISO/IEC 42001 · NIST AI RMF · EU AI Act · SOC · ISO · FedRAMP · CMMC
Our role
Advisory, readiness and testing

Assurance speaks to someone not in the room.

We help you choose the assurance your buyers recognize, build controls that operate, and test the AI they describe.

For SOC reports and ISO certificates, we take you from readiness to a clean result with the CPA firm or certification body that signs it.

AI governance & security

Governance that knows how models break.

Frameworks that decide how AI is governed, and testing that shows whether safeguards hold.

NIST AI RMF 1.0

NIST AI Risk Management Framework

Govern, Map, Measure and Manage, plus the Generative AI Profile (NIST AI 600-1). Voluntary, and increasingly the vocabulary US buyers expect.

  • Current and target profiles
  • Generative AI risks applied to your use cases
  • Integrated with ISO/IEC 42001
Regulation (EU) 2024/1689

EU AI Act Readiness

The Digital Omnibus moved most high-risk obligations to 2 December 2027 (Annex III) and 2 August 2028 (Annex I). Classification decides what applies to you.

  • Inventory and risk classification
  • Provider and deployer role analysis
  • Obligation mapping and plan
Read: the AI Act after the Digital Omnibus
AI agent standard

AIUC-1 Readiness

Control requirements plus recurring adversarial testing for AI agents, revised quarterly.

  • Gap assessment against current requirements
  • Pre-certification adversarial testing
Adversarial testing

AI Red Teaming

Adversarial testing of models, agents and the applications around them, under agreed rules of engagement.

  • Direct and indirect prompt injection
  • Data leakage and unsafe tool use
  • Mapped to OWASP LLM Top 10 and MITRE ATLAS
See synthetic red-team findings
Application security

LLM & AI Application Security

Most AI vulnerabilities are architecture decisions made before the first prompt.

  • RAG, agent and plugin threat models
  • Vector-store tenancy and entitlements
  • Model and dataset supply chain
Read: prompt injection is an architecture problem
Core service

ISO/IEC 42001 AI Management Systems

The management system that ties these five services together and carries you to certification.

Explore ISO/IEC 42001
Defense supply chain

CMMC and NIST SP 800-171.

Readiness for contractors handling controlled unclassified information, with the evidence a C3PAO needs to certify you at Level 2.

CMMC / NIST SP 800-171

CMMC & NIST SP 800-171 Readiness

Level 2 aligns to the 110 requirements of NIST SP 800-171 Rev. 2. CMMC clauses began appearing in DoD contracts on 10 November 2025, and C3PAO certification becomes the default for Level 2 contracts from 10 November 2026.

  • CUI scoping and enclave design
  • Gap assessment with SPRS scoring
  • SSP, POA&M and mock assessment

Level 2 certification assessments are performed by authorized C3PAOs; Level 3 by the government.

Choosing the right assurance

Which report does your buyer want?

The fastest route to a deal is the assurance your customer recognizes.

If the question comes from…They usually wantIssued by
US enterprise security reviewsSOC 2 Type IILicensed CPA firm
Customers’ financial auditorsSOC 1 Type IILicensed CPA firm
International procurementISO/IEC 27001Accredited certification body
Buyers asking about AIISO/IEC 42001Accredited certification body
US federal agenciesFedRAMPFedRAMP, on a 3PAO assessment
US defense contractorsCMMCSelf-assessment, a C3PAO or the government
US healthcareHITRUSTHITRUST, via an Authorized External Assessor

Need more than one? One evidence model, several frameworks

FAQ

Questions we are asked.

Which assessments does MarkTillson perform itself?

For SOC reports and ISO certificates, we take you from readiness to a clean result with the CPA firm or certification body that signs it.

We need SOC 2 now and an AI answer soon. Where do we start?

With a single control set. Access, change and vendor controls underpin SOC 2, ISO/IEC 27001 and 42001 alike; then we add AI-specific elements.

Is AI red teaming different from a penetration test?

It adds model behavior: prompt injection, jailbreaks, leakage and unsafe agent actions. We often run both, because the worst AI findings combine a model weakness with an over-privileged integration.

Has the EU AI Act been delayed?

Partly. Most high-risk obligations moved to December 2027 and August 2028; prohibitions, AI literacy, GPAI and Article 50 duties were not deferred the same way.

Can you help a defense contractor without a cloud product?

Yes. CMMC readiness focuses on where CUI lives. Isolating it in an enclave often cuts cost substantially.

Speak with the practice

Tell us who is asking for assurance.

A customer, a regulator, an agency or your board. A senior practitioner will help you choose the shortest credible route.