Home / Governance & risk

Governance that operates, not governance that sits in a drive.

Every framework below was built to be run by the people who'd be audited on it. Each starts from what the organisation already does and maps it to NIST AI RMF, the EU AI Act, ISO/IEC 42001 and the local regulator.

GCB Bank PLC · Automation Engineering Lead · 2025

A 4,000-person bank's first AI governance framework

The bank was adopting GenAI and automation with no AI governance at all. I designed the framework that every AI initiative now passes through.

0 → 1

AI governance infrastructure, bank-wide

Read the case study

Context

A Ghanaian bank with 4,000+ staff, a fast-growing automation pipeline and a first GenAI deployment on the way. No AI policy, no risk tiering, no named owners.

What I owned

Framework design end to end, and the governance gates inside the Centre of Excellence operating model.

Approach

  • Aligned the structure to ISO/IEC 42001 so it could later be certified
  • Put Risk, Compliance and Internal Control gates at intake, not at go-live
  • Risk-tiered every use case before build

Execution

  • Ethical AI principles and compliance standards adopted bank-wide
  • Intake and assessment framework embedded in the CoE pipeline
  • Phased roadmap rule: high-risk agentic features deferred until controls exist

Results

4,000+staff in scope
8business units under one gated pipeline
CoE standardrisk-tiered phased roadmap adopted

AlgoViva · Senior Practitioner · 2026

Enterprise AI policy suites for mid-sized clients

Clients needed policies their teams could follow and their regulators would recognise — across several jurisdictions at once.

5

policy domains per suite

Read the case study

Context

Mid-sized organisations in Africa adopting AI tools faster than their policies could keep up, facing both global standards and national rules.

What I owned

Co-creation and implementation of the policy suite, plus leadership training ahead of rollout.

Approach

Every control mapped to its source: NIST AI RMF, the EU AI Act, ISO/IEC 42001 and country-specific regulation. One control, many citations — so one policy answers several regimes.

Execution

  • Governance structure, risk management, accountability, transparency and vendor-risk policies
  • AI safety training for teams and leadership adapted from BlueDot Impact and Atlas AI Safety curricula

Results

5policy domains
4frameworks mapped per control

Ghanaian healthtech AI lab · 2026

A three-phase governance programme for a healthtech AI lab

A health AI startup needed to show partners and regulators it could be trusted with clinical data and model decisions.

36

governance artefacts delivered

Read the case study

Context

Early-stage healthtech building AI on sensitive health data, with partners asking for evidence of governance before signing.

What I owned

The full programme across three phases, from gap analysis to the complete artefact set.

Approach

Built against ISO/IEC 42001 and NIST AI RMF together, so each artefact satisfies both. Linked every artefact to the others it depends on.

Execution

  • 36 artefacts: policies, procedures, registers, assessments and templates
  • An interactive map showing how every artefact connects, so the team can see what breaks if one is skipped

Results

36artefacts
3phases
2frameworks covered by each artefact

AlgoViva · 2026

Which AI tools can staff actually use?

Employees were already using AI apps. I turned an open question into an approved list and clear rules.

211

AI apps benchmarked

Read the case study

Context

Shadow AI use with no way to tell safe tools from risky ones.

What I owned

Evaluation method, approved-tool list and the usage guardrails.

Approach

Benchmarked candidate tools against the VerifyWise AI App Trust & Transparency Index: 211 apps scored across seven data-governance domains.

Execution

  • Approved-tool list with conditions of use
  • Internal guardrails on what data may go into which tool

Results

211apps assessed
7data-governance domains

Advans Group · Fractional AI Governance Advisor · 2026

AI charter review before a customer-facing rollout

A microfinance group wanted its AI charter stress-tested before customers saw the product.

Pre-launch

liability gaps documented

Read the case study

Context

A draft AI charter about to govern a customer-facing deployment.

What I owned

Independent review and written recommendations.

Outcome

Documented recommendations on liability exposure and responsible deployment, delivered ahead of the customer rollout.

Need a framework your auditors will accept?

Start with a 30-minute call. Bring your current policies, or nothing at all.

Book a callTry the readiness check