Strategy & Ownership
Does an AI initiative have a business owner who carries the outcome, or only a sponsor who approved the budget?
- Business case per use case
- Named accountable owner
- Portfolio prioritisation method
- Board-level reporting line
Consulting · AI Readiness Assessment
Know where you stand before you commit a budget.
Most enterprise AI programmes do not fail on the model. They fail on data nobody owns, integrations nobody scoped and governance nobody agreed. We assess all six of those dimensions honestly, then hand you a sequenced roadmap your board can actually fund.
- Engagement length
- 4–6 weeks
- Scales with entity count and data access
- Delivery model
- On-site + remote
- Bandar Seri Begawan and Berakas teams
- Track record
- 20+ years
- Serving Brunei enterprises since 2006
- Vendor stance
- Independent
- Deliverables are yours to take anywhere
Why it matters
The expensive failure is the one you fund twice.
The pattern is consistent enough to be predictable. A capable team runs a convincing pilot. Leadership approves a programme. Six months later the programme is stalled — not because the technology disappointed, but because the data it needed was owned by three departments who disagreed about its definition, and because nobody had decided what the AI was permitted to do without a human.
By then the organisation has spent a budget cycle and acquired a reputational problem: people now believe AI does not work here. The second attempt is harder to fund than the first.
A readiness assessment is deliberately unglamorous work done in the cheapest possible order. It establishes what is true about your data, systems, governance and people before a commitment is made, so that the sequence you fund is one that can actually complete.
- Findings are evidenced, not asserted — we show you what we looked at
- Gaps are named plainly, including the ones that delay your timeline
- The roadmap is costed, sequenced and owned before anyone writes code
The framework
Five levels. Most organisations sit between two of them.
Maturity is assessed per dimension, not as a single company-wide score — it is entirely normal to be operational on knowledge access and ad hoc on governance. Select a level to see what it looks like from the inside.
Level 01 · Curiosity
Ad hoc
Individuals experiment with public AI tools. Nothing is documented, nothing is governed, and no one owns the outcome.
What we typically see
- Personal tool subscriptions
- No acceptable-use position
- Data leaving the organisation unrecorded
- Enthusiasm concentrated in one or two people
The next move
Establish an acceptable-use position and inventory what is already in use before restricting anything.
Level 02 · Pilots
Exploring
Departments run isolated pilots that demo well. They stall at the point where they would need production data or a real owner.
What we typically see
- Several pilots, no production users
- Pilots blocked on data access
- No agreed success measures
- Budget approved per experiment
The next move
Score the pilot portfolio on value against effort and kill the ones that cannot reach production.
Level 03 · In production
Operational
At least one AI capability serves a real workflow, with named ownership and a defined review point when it behaves unexpectedly.
What we typically see
- Live capability with real users
- Named business owner
- Human review on consequential steps
- Basic usage monitoring
The next move
Formalise the governance you are already doing informally, then extend the pattern to the next workflow.
Level 04 · Platform
Integrated
AI capability sits on shared foundations — one identity model, one permission model, one audit trail — rather than being rebuilt per department.
What we typically see
- Shared retrieval and permission layer
- Reusable tool and integration catalogue
- Consistent audit trail across capabilities
- Central cost visibility
The next move
Shift the conversation from individual use cases to platform economics and portfolio prioritisation.
Level 05 · Assured
Governed at scale
AI is a governed operating capability: risk-classified, evidenced, measured against business outcomes and defensible to an auditor or a ministry.
What we typically see
- Risk classification per use case
- Evidence generated as a by-product
- Outcome measurement in management reporting
- Formal change control on models and prompts
The next move
Maintain the assurance posture and use it commercially — in tenders, in due diligence, with regulators.
What we assess
Six dimensions, each with a question you cannot bluff.
Data Readiness
Can the data an AI capability would depend on be trusted, accessed and permissioned without a project of its own?
- Source inventory & ownership
- Quality and completeness baseline
- Access and classification model
- Retention and residency constraints
Technology & Integration
Can systems expose what an AI capability needs to read and write, or would every integration be bespoke?
- API surface per core system
- Identity and single sign-on
- Event and audit plumbing
- Environment and release discipline
Governance & Risk
Where must a human stay in the loop, what may an AI feature never decide alone, and how would you evidence that later?
- Human-in-the-loop boundaries
- Risk classification approach
- Acceptable-use position
- Audit and traceability requirements
Infrastructure & Sovereignty
Do your deployment options actually satisfy the data-residency and sovereignty constraints you operate under?
- Cloud, hybrid or on-premises fit
- Model hosting options
- Cost and capacity envelope
- Resilience and continuity
People & Change
Whose day changes, who has to be brought along, and who currently has the skills to run what you are proposing?
- Role-level impact mapping
- Capability and skills baseline
- Adoption and training plan
- Communication sequence
Current vs future state
The gap, written down where everyone can argue with it.
Agreement on the current state is usually the hardest part of the engagement, and the most valuable output of it.
| Area | Where most organisations are | Where the roadmap takes them |
|---|---|---|
| Decision support | Reports compiled manually, disagreeing depending on who ran them | One agreed definition per measure, traceable to the transactions behind it |
| Knowledge access | Answers depend on asking the right colleague before they go on leave | Permission-aware retrieval over approved sources, cited back to the document |
| Routine processing | Skilled staff re-keying between systems that do not talk to each other | Assisted preparation with a human approving the consequential step |
| Governance | AI use is unmapped, so risk is unquantified and unevidenced | Risk-classified use cases with an audit trail produced as a by-product |
| Cost visibility | Departmental AI subscriptions invisible to finance | Central cost and usage visibility against a prioritised portfolio |
The scorecard
A score is only useful if you can see how it was reached.
Each dimension carries a level, the evidence behind that level, and the specific gap that holds it back. There is no composite vanity number, because a single score invites the wrong conversation — the useful discussion is always about the weakest dimension.
Scores are agreed with you in the readout, not delivered as a verdict. Where we disagree with a process owner about a level, that disagreement is recorded rather than resolved by us.
Deliverables
Six artefacts you keep.
Yours to take to any vendor, including one that is not us. The engagement is priced as advisory work, not as a route into an implementation contract.
- 01
Maturity Assessment Report
Your position on the five-level ladder per dimension, with the evidence behind each score and the gaps stated plainly — including where the honest answer is "not yet".
- 02
Use-Case Portfolio
Candidate use cases scored on business value against implementation effort, each with its data dependencies and a build, buy or defer recommendation.
- 03
Sequenced Adoption Roadmap
A phased plan with dependencies made explicit — what must land first, what can run in parallel, what waits — and a named owner per phase.
- 04
Governance Guardrails
Human-in-the-loop points, approval boundaries, acceptable-use position and the audit-trail requirements needed to evidence all of it later.
- 05
Business Case
Cost, the assumptions behind it, the measures that will tell you whether it worked, and the review checkpoints where you can stop.
- 06
Executive Readout
A leadership session that walks the findings, the recommendation and the trade-offs — and answers the question the board will ask.
Business outcomes
What the assessment changes.
Stated as outcomes rather than percentages. We do not publish figures from other clients' programmes as if they were a forecast for yours.
Faster AI adoption
Sequencing removes the stall points before they consume a budget cycle.
Better funding decisions
Value-versus-effort scoring puts money behind the defensible cases, not the loudest ones.
Reduced compliance risk
Guardrails and evidence requirements are defined up front rather than retrofitted under audit.
Enterprise scalability
Shared foundations mean the second use case costs a fraction of the first.
Industries
Sector context changes the answer.
A ministry's residency constraints and a distributor's margin pressure lead to different roadmaps from the same maturity score.
FAQ
What buyers ask before engaging.
How is this different from a vendor proof of concept?
A proof of concept answers "can this technology work?". This answers "should we fund it, in what order, and what has to be true first?". The output is a set of decisions and documents you own, not a demo environment that expires. We would like to build what follows, but the engagement is not structured to require it.
What if the assessment concludes we are not ready?
Then we say so, and the roadmap starts with foundations — process documentation, data remediation, integration groundwork — instead of a model. That answer has saved clients considerably more than the engagement fee, and it is the reason the assessment is scoped and priced separately from any build.
How much of our team's time does it take?
Expect a half-day workshop with leadership, then roughly two to three hours each from the process owners in scope, plus access to sample data and system documentation. We work around operational schedules and can run sessions across your sites.
Do you only recommend your own AI ERP platform?
No. Where our platform genuinely fits we will say so and explain why; where an off-the-shelf product or your existing stack is the better answer, we will say that instead. The scoring criteria are shared with you, so you can check the reasoning rather than take it on trust.
Can the output satisfy a tender or regulator question about AI governance?
The guardrails deliverable covers acceptable use, human-in-the-loop boundaries and audit-trail requirements, which addresses most tender questions. For a full ISO/IEC 42001-aligned management system, pair this with our Cybersecurity & AI Governance engagement.
Does this cover data residency and sovereignty?
Yes — it is one of the six dimensions. We assess whether your deployment options actually satisfy the residency constraints you operate under, including on-premises and hybrid hosting of models where that is the requirement.
Get started
Start with an honest assessment.
A short conversation is usually enough to tell whether you need the full six-dimension assessment or just a second opinion on a decision you have already taken.
