AI Governance Framework
What needs to be governed.
A catalogue of controls with owners and evidence, mapped from ISO/IEC 42001, the EU AI Act and NIST AI RMF, with a four-level maturity scale. It answers what has to be in place.
Methodology
From readiness to realized value.
AGORA's Engagement Delivery Framework is the structured way to move a data and AI initiative from initial qualification through implementation and sustained value.
Why AGORA exists
Every professional services organisation that grows faces the same problem: the quality of the deliverable becomes a function of who delivers it, not of how it is delivered. Projects that go well are credited to individual talent; those that go badly, to client circumstances. Neither explanation lets anyone learn.
The framework exists so the outcome does not depend on who touched the project. It defines the what and the why of each phase; the how stays with the team.
The five phases
They are not extra phases, nor stages that happen at the end. They are lenses applied across all five phases, each with a named owner. Omitting a single one is the most common cause of technically correct projects that fail in production.
- 01
Qualify
Should this engagement happen?
Validate outcome, sponsorship, feasibility and material risk before committing resources.
→ An explicit decision to proceed or not, with its reasons written down.
- 02
Discover
What is the real problem?
Replace assumptions with evidence across business, stakeholders, data and technology.
→ An agreed, measurable definition of success — not a wish list.
- 03
Design
What is the executable plan?
Turn evidence into priorities, architecture, risk treatment and a delivery plan.
→ A plan that can be started without structural ambiguity.
- 04
Deliver
Is the solution producing the intended outcome?
Execute in short iterations while validating functionality, quality, adoption and operational readiness.
→ The solution in production, and the client's team able to operate it.
- 05
Sustain
Is the value holding?
Verify that adoption, operational performance and value persist after go-live.
→ Evidence that what was promised materialised — or that it did not, while it can still be corrected.
Cross all five phases
- Security & ComplianceThat the solution respects the client's security standards and the applicable regulatory framework, verified continuously rather than at the end.
- Responsible AIFairness, transparency, explainability and human oversight in every engagement with an AI component, from classical models to systems that decide and execute.
- Data GovernanceInventory, classification, lineage, ownership and quality of the data the solution touches. It belongs to the delivery team; it is not outsourced.
- Risk ManagementA living register from the first phase to the last, covering technical, commercial, regulatory and organisational risk, each with an owner and a plan.
- Knowledge ManagementThat what was learned is not lost to team turnover or trapped in one person's head. Decisions documented where they can be found later.
Independent quality review at critical decision points
Moving from one phase to the next is neither automatic nor left to whoever runs the project. It rests on verifiable criteria, documented evidence and an independent senior review.
- Evidence
- What is claimed at the close of a phase has to be backed by something another person can review.
- Independent review
- Reviewed by someone who did not produce it. Whoever did the work is not who decides whether it is sound.
- Decision
- Proceed, correct or stop. All three are legitimate outcomes and are recorded with their reason.
Two systems, two different questions
AGORA works with two proprietary frameworks that do not overlap: one defines what needs to be governed, the other how the work gets delivered. Confusing them is why many firms end up with a policy nobody executes, or a delivery nobody can audit.
Engagement Delivery Framework
How the work gets delivered.
Five phases each with a central question, five dimensions crossing them, and independent review at every crossing. It answers how you get from an idea to sustained value.
The starting point is always the same
Before designing anything you need to know where the organisation stands. That is the first phase, and also the product you can start with.