What it takes to deliver this
Estimate the full offer as an integrated consulting, learning, change, data and product-engineering engagement. A small training team cannot absorb the engineering, integrations and documentation without explicit capacity.
- 01
Align leadership and functions before launch
IT and HR can both claim AI fluency. Agree ownership at C-suite level before anything goes live.
- 02
Governance comes before learning
People need an AI use policy they understand before they are trained to use the tools.
- 03
Build a common foundation first
Do not assume everyone starts with the same AI knowledge, not even the champions.
- 04
Be deliberate about the pace of change
Enterprise products take time and iteration. Check that the client can absorb continuous change.
- 05
Plan for issues specific to AI products
Token use policy, IP, stricter security and access. Multi-region builds need strong technical oversight.
| Area | First delivery | Next engagements |
|---|---|---|
| Sales process | A conceptual pitch, no established story, several pricing and scoping rounds. | Working demos and decks, and validated reference figures. |
| Starting point | Little insight into what leaders wanted from AI. Learning material and tools built from scratch. Governance and access issues. | A clear view of what leaders want and how to coordinate functions. A baseline for learning and tools. Governance and access addressed in the first weeks. |
| Team | Roles found during delivery. | The same roles, plus an AI tech lead. Leaner teams can deliver tailored products, because the baseline assets exist. |
| Launch | Limited launch experience; development timelines underestimated. | A proven implementation approach and realistic timelines. |
Engagement lead
Owns the client relationship, ambition, scope and steering.
Executive program lead
Runs pre-meets, tailors and facilitates the executive sessions, writes the 90-day plan.
Learning designers
Persona analysis, foundations and persona paths, facilitator material.
Change and communications
Why story, communications toolkit, champions and manager activation.
Platform engineers
Engagement Hub on SharePoint, CAiRA (the Learning Agent) in Copilot Studio, Power Automate flows.
Data and reporting engineers
Cockpit in Power BI: data sources, refresh, metric dictionary, access model.
Product owner
Release priorities, acceptance, backlog and handover of the products.
AI tech lead (new, proposed)
Technical oversight of the Hub, Agent and Cockpit builds, across regions and environments.
Clients buy three phases (Foundation, Scaling, Embedded). Each phase runs through six stages, and each stage ends with a gate passed together with the client.
- 01
Shape
Agree ambition, scope, audiences, value hypotheses and ownership.
Gate: Sponsors and delivery owners agree what is included, excluded and owned by the client.
- 02
Design
Design learning, leadership, communications, products, measurement and support as one experience.
Gate: Content, technical, data and operational designs are reviewed together.
- 03
Build
Produce the agreed assets and test them in the client's environment.
Gate: Agreed functional, content, access and operational criteria are met.
- 04
Activate
Launch to the agreed audience and reinforce practical use.
Gate: Access, support, content and data refresh work for the launch population.
- 05
Stabilize
Resolve issues, set up recurring operations and improve the experience using evidence.
Gate: Support is handed over and a scale recommendation is agreed.
- 06
Scale
Extend to new audiences, products or use cases once readiness and ownership are in place.
Gate: Each extension has an owner, a baseline and the capacity to run it.
Where effort was underestimated
- Persona analysis needs a named owner and an explicit depth.
- LMS upload responsibilities must be explicit in the SOW.
- LMS and gamification data have to be exposed by the client.
- Dashboard and data-connection work need dedicated capacity.
- The hub, automation and the agent are real engineering components.
- SDLC, CI/CD, deployment pipelines and documentation can consume substantial effort.
Define for every module
- Deliverable and intended use
- Quantity and depth
- Client inputs
- Our responsibilities and the client's
- Acceptance evidence
- Exclusions
- Change-control triggers
- Handover and ongoing ownership
What the client brings
- Approved content and policies
- Business and persona representatives
- Tool access for learners and facilitators
- Development, test and production environments
- Identity and access decisions
- Learning and gamification data
- Reporting data and business definitions
- Approval availability
- LMS publishing capacity
- Content and support owners
- Release and security-review processes
Learning material
- Executive upskilling: hands-on discovery and one-to-one upskilling
- AI fundamentals: decks for a 3-hour e-learning course
- Advanced Copilot: three levels of decks and facilitator guides for 15 personas
Frameworks
- CXO guide: one page on how to use AI well
- AI certification: belt levels and achievement requirements
- Gamification: a points framework for an organization-wide leaderboard
- Behavior change: a framework for each certification level
Toolkit
- Leader support: an AI implementation guide for leaders
- Engagement handbook: hackathon, vibe-coding and prompt-a-thon kits
- Manager activation: steps to strengthen team AI adoption
Reusable does not mean plug-and-play
Reuse from day one
Learning content, frameworks, playbooks, engagement kits, product designs and wireframes
Scope per client
Architecture, integration, testing, governance, security and token policy, IT involvement
What reuse saves on the next engagement
Estimates per asset from the first delivery. Not measured savings: do not quote them as a commitment.
- Engagement HubAbout 30% fewer development hoursIllustrativeThe demo version needs adapting to client requirements and branding, not building from scratch.
- CAiRA, the Learning AgentAbout 25% faster to buildIllustrativeThe demo version and its documentation exist.
- Value & Adoption CockpitAbout 30% faster to buildIllustrativeThe demo version, metric definitions and handbook exist.
- AI fundamentalsAbout 15% faster to buildIllustrativeBaseline content exists, so the same team finishes faster.
- Other non-technical assetsAbout 20% faster to deliverIllustrativeToolkits, playbooks and frameworks are reused and tailored.
Asset inventory
| Asset | What it is | Status | Note |
|---|---|---|---|
| Communications and activation toolkit | Communication tools, engagement activities, manager and leader activation | Confirm before reuse | Branded and generic versions exist. Check completeness and reuse rights. |
| CAiRA requirements and design | Sanitized BRD and TDD templates | Confirm before reuse | Documentation, not a deployable package. |
| Learning content and gamification | Foundations, persona modules, belt and proficiency framework | Confirm before reuse | Substantial base. Client-specific reuse needs validation. |
| Executive program | Four-session curriculum, pre-meet guide, continuation formats | Available | Reusable pattern; tailored per client. |
| Value & Adoption Cockpit | 18 metrics across literacy, adoption and value; operational handbook; demo with synthetic data | Confirm before reuse | The demo on this site uses synthetic data. Check which release is reusable before scoping. |
| Delivery accelerators | Persona Needs Analyzer (PNA) and LearnDex, the learning content indexer | Confirm before reuse | Built on Copilot Studio, for internal use. Keep out of client-facing material. |
| Capgemini demonstrator | CAiRA, Hub and Cockpit on Capgemini's own platforms, with backend documentation | Available | Demo-ready for client conversations (VPD deck V3.1). Adapt to client requirements and branding before use. |
Before reusing any asset with a client: confirm sanitization, reuse rights, version and support owner.
The internal portal is hidden in client view. Switch to internal view in the top bar.