The Sequenced Roadmap: 12 to 36 Months of AI Enablement

Putting it all together

You now have the framing (Module 1), the augmentation/redesign diagnostic (Module 2), the workflow redesign framework (Module 3), the decision rights model (Module 4), the data layer perspective (Module 5), and the talent and operating model implications (Module 6). This final module is about how to sequence all of that into a programme that actually lands inside a real organisation, over a real timeline, with real stakeholders.

The single most common reason ambitious AI enablement programmes fail is not technology and not capability — it is sequencing. Trying to do too much in parallel. Scoping the data work as a separate stream that has to finish before anything else can start. Tackling the politically hardest workflow first. Underestimating how much organisational attention a real redesign consumes. Losing stakeholder patience before any phase delivers a visible outcome.

This module gives you the sequencing pattern we use to avoid those mistakes.

A three-phase, 12–36 month arc

Most enabled-by-design programmes we run fit into a three-phase arc.

Phase 1 — Prove the pattern (Months 0–6)

Goal: redesign one priority workflow, end-to-end, including its data layer, decision rights, and governance. Prove that the operating pattern works in your context. Build the playbook and the institutional muscle.

Scope: one workflow. Not three, not the whole function — one. The choice of which workflow matters enormously. Pick one that meets all of these criteria:

• Material to the business (real revenue, real risk, real customer experience)
• Tractable in its data requirements (you can fix the data layer in 4–6 months for this specific workflow)
• Owned by an executive sponsor who will hold the political space
• Bounded enough that you can deliver a defensible outcome in 6 months
• Visible enough that the rest of the organisation will see it land

Outputs:

• One redesigned, in-production workflow that demonstrably outperforms the legacy version
• The data layer slice required to support it
• Decision rights matrix and governance framework
• A trained team (workflow designer, system supervisor, exception handlers, feedback curator)
• A reference playbook for Phase 2

The unlock: when Phase 1 delivers, you have proof that AI enablement is real in your organisation. The political capital that produces is what makes Phase 2 possible.

Phase 2 — Scale the pattern (Months 6–18)

Goal: apply the playbook to 3–5 additional priority workflows in adjacent domains. Scale the data layer outwards from the Phase 1 anchor. Begin to make the operating model changes structural rather than experimental.

Scope: workflows that share some of the data, governance, or operational context with the Phase 1 workflow — so you can re-use as much of the foundation as possible. Resist the temptation to jump into politically harder territory just because Phase 1 was successful; the muscle isn't yet strong enough.

Outputs:

• 3–5 additional workflows redesigned and in production
• The data layer extended into adjacent domains
• Cross-workflow patterns codified (decision rights, governance, role design)
• The operating model changes start to become permanent (new metrics, new roles, new career paths)
• Executive narrative that AI enablement is now the way the function operates

Phase 3 — Embed and compound (Months 18–36)

Goal: AI enablement stops being a programme and starts being how the organisation operates by default. New workflows are designed AI-native from the start. The data flywheel begins to compound.

Scope: the rest of the function, plus the structural pieces that have been waiting for political readiness — the harder workflows, the cross-team dependencies, the legacy systems that needed to be replaced before they could be redesigned.

Outputs:

• Most workflows in the function operate AI-native by default
• The data layer is a reusable platform, not a per-workflow rebuild
• New work starts in the AI-native pattern; no new augmentation projects are commissioned
• The talent model is structurally different from where you started
• The organisation has compounding advantage that competitors will struggle to close

The compounding audit

One specific workstream that should run alongside Phase 1: the compounding audit of your existing AI pilots.

Most enterprises start an enablement programme with a portfolio of existing AI initiatives — chatbots, copilots, internal tools, vendor platforms, experiments. These are not the enemy. They are data.

The audit asks one question of each: under the new operating model, will this initiative compound or will it stay local? The answer determines what happens next:

• Compounds → migrate it into the new framework, with proper decision rights, governance, and integration into the broader workflow.
• Doesn't compound → kill it deliberately, with a documented rationale, so you don't end up paying for legacy initiatives that aren't going anywhere.
• Unclear → put it on a 6-month watch list with explicit success criteria, then decide.

This audit is politically delicate, because every existing initiative has a sponsor and a budget line. But it is essential. Without it, you end up running the new operating model on top of the old one, which is the worst of both worlds.

Governance and the regulator

For our financial services and regulated-industry clients, the governance work runs alongside the technical work from day one. By the end of Phase 1, you should be able to:

• Show a regulator the decision rights matrix for the redesigned workflow
• Demonstrate the override rate and how you monitor it
• Walk them through the model risk framework, including drift detection and rollback procedures
• Map the workflow to the relevant regulatory expectations (EU AI Act, PRA SS1/23, FCA SYSC, DORA)
• Show the audit trail for any individual decision

If you can do that for one workflow at the end of Phase 1, you can replicate the pattern for every subsequent workflow. If you can't, you have governance debt that will catch up with you in Phase 2.

The right way to think about governance in AI enablement is not "compliance overhead" but "accelerator." Done well, governance is what unblocks AI deployment by removing the institutional friction that normally kills enterprise initiatives. Done badly, it is the brake everyone fears. The choice is yours.

Stakeholder management is the real work

The technical work in AI enablement is hard but solvable. The data work is harder but solvable. The role design is harder still but solvable. The thing that consistently kills programmes is stakeholder management — and specifically, the inability to hold political space through the periods when no visible outcome has yet landed.

Phase 1 takes 6 months. For most of those 6 months, nothing visible is happening. The data work is invisible. The workflow design is invisible. The role redesign is contentious. The team is heads-down on something the rest of the organisation doesn't yet understand. This is the moment when CFOs ask hard questions, when steering committees suggest scope cuts, when other priorities try to grab the team's attention.

The executive sponsor's job is to hold that political space. Not to manage the technical work. Not to attend every steering committee. To say, in the rooms where it matters, "we are going to deliver this, and I have authorised it, and I need you to give us the time." Without that, no programme survives.

If you take one thing from this final module, take this: the technical playbook is now written, and it works. The organisational playbook is harder, takes longer, and is the actual job.

Ready for the exam

You have now covered the seven modules of the AI Enablement for Operations Leaders course:

1. The production function is changing
2. Augmentation vs redesign — the diagnostic
3. The workflow redesign framework
4. Decision rights in an AI-native operating model
5. The data layer — the constraint that determines everything
6. Talent, roles, and the new operating model
7. The sequenced roadmap

The final assessment is 25 multiple-choice questions covering the full course. Pass mark 70%. On completion you will earn your AI Enablement Practitioner certification from Insight Centric.

Good luck — and when you are ready to apply this framework to a real programme in your organisation, our AI Enablement service is built around exactly this model.