CELA demonstration script — presenter companion
From task visibility to governed execution
Planner as the commitment layer, Cowork as the coordination layer
The challenge for CELA programme managers is not only visibility of work. It is trusted coordination across legal, privacy, policy, product, security, communications, and executive stakeholders. Planner holds the commitments. Cowork accelerates the coordination. Human validation, source traceability, and approval checkpoints preserve accountability at every step.
Run of show
| Act | What it shows | Underlying message |
|---|---|---|
| Act 1 | Planner board for the readiness programme | Planner is the system of record for cross-functional commitments |
| Act 2 | Cowork reviews tasks, surfaces risk and bottlenecks | Cowork accelerates coordination — it does not replace PM judgment |
| Act 3 | Human validation checkpoint before anything moves | AI supports the PM; it does not become the accountable owner |
| Act 4 | Source validation view linking output to evidence | Trust comes from traceability, not polished output |
| Act 5 | Stakeholder-specific communication drafts | The right communication is the appropriate one, not the fullest one |
| Act 6 | Executive decision memo, drafted post-validation | Executives get decision clarity, not a task dump |
| Act 7 | Calendar checked, review meeting scheduled, Teams messages drafted and sent to task owners | Coordination becomes real action — still gated by PM approval before anything sends |
| Act 8 | Approval checklist and final sign-off path | Governance is explicit, not assumed |
| Act 9 | Pilot measurement framework | The model must prove improvement, not just look impressive |
Governance architecture
Every act in this demo maps onto a single governance chain. Watching where each act sits on the spine is how the audience sees that acceleration never skips accountability.
“Think of Planner as the commitment layer and Cowork as the coordination layer. The commitment layer never moves without a human. The coordination layer accelerates everything up to — but not past — the point of human judgment.”Presenter framing
Planner board setup
This is the live board Cowork reads from throughout the demo — the EU AI Act Conformity Readiness Programme, shown as a Planner board with 10 workstream buckets and 11 commitments. Every prompt you run below is grounded in exactly these tasks, owners, dates and evidence — nothing is invented.
EU AI Act Conformity Readiness Programme
Planner · Board view
Regulatory Interpretation
1Map relevant EU AI Act obligations
Map applicable EU AI Act obligations to product portfolio
Legal Review
1Complete legal interpretation note
Complete legal interpretation note on high-risk classification
Blocks executive memo and public policy position.
Privacy & Data Protection
1Assess GDPR and privacy implications
Assess GDPR overlap and data-processing implications
Product Readiness
1Prepare product impact summary
Prepare product impact summary for affected features
Security Review
1Review security control evidence
Review security control evidence against conformity requirements
Policy & External Affairs
1Draft public policy position
Draft public policy position statement
Blocked pending final legal interpretation note.
Executive Governance
2Prepare executive decision memo
Prepare executive decision memo for governance review
Cannot be finalised until legal and privacy inputs are validated.
Validate Cowork risk scan before executive use
Explicit governance checkpoint — accountability remains with the PM.
Communications Readiness
1Create stakeholder communication plan
Create stakeholder communication plan
Implementation Tracking
1Confirm regional stakeholder impacts
Confirm regional stakeholder impacts (EU markets)
No owner assigned — coordination gap.
Evidence & Audit Trail
1Attach final approved evidence pack
Attach final approved evidence pack to audit trail
"Notice what the board already tells us before any AI runs: a legal note is overdue, a policy task is blocked behind it, and one regional task has no owner at all. Cowork's job is to make those coordination gaps impossible to miss — not to decide what to do about them."Presenter framing
Act 1
Establish the programme control surface
Show Planner as the structured source of truth for programme work — before Cowork is introduced at all.
- —Buckets, task owners, due dates, labels, priority
- —Checklist items, attachments, linked documents
- —Comments or activity history, where available
“Let us begin where CELA programme management actually begins: with commitments. This board represents a live cross-functional readiness programme. The important point is that Planner gives us structured visibility. But visibility alone does not move the programme. The PM still has to interpret risk, chase dependencies, prepare decision material, and coordinate stakeholders. That is where Cowork becomes useful.”Presenter script
Act 2
Cowork programme risk scan
Cowork analyses the Planner task set and prepares a risk-based programme view — organising evidence, not making legal judgments.
“Notice the instruction: Cowork is not being asked to make the legal decision. It is being asked to organise the programme evidence, identify coordination risks, and recommend where human attention is needed.”Presenter script
# Role Act as a CELA Programme Management assistant supporting an EU AI Act Conformity Readiness Programme. # Context Use the Planner board as the system of record for programme commitments. Review tasks across Legal, Privacy, Product, Security, Policy, Communications, and Executive Governance workstreams. # Objective Prepare a programme risk scan for the CELA PM team. # Analyse Identify: - overdue tasks - tasks due soon - missing owners - blocked tasks - approval bottlenecks - workstreams with insufficient evidence - dependencies that may affect executive sign-off - actions requiring Legal, Privacy, Security, Product, or Policy input # Output ## 1. Programme Health Classify the programme as: On Track / At Risk / Off Track. Explain the classification based only on available Planner and Microsoft 365 context. ## 2. Critical Risks For each risk: description, affected workstream, related Planner task, likely implication, recommended next action, recommended human owner for validation. ## 3. Approval Bottlenecks Identify approvals or reviews that may delay executive sign-off. ## 4. Immediate PM Actions Recommend actions for the PM team to take next. ## 5. Information Gaps Clearly state any information that is missing or not available. Do not invent owners, approvals, dates, legal conclusions, or regulatory interpretations.
Act 3 — the pivotal act for this audience
Governance trust layer
Answers the exact question Legal, Privacy, and Security will ask: how do we trust the AI-generated synthesis?
On-screen: a Planner task titled "Validate Cowork Risk Scan Before Executive Use."
“Before any recommendation goes to executives, the workflow introduces a validation step. Cowork has generated a programme risk view, but accountability remains with the named human owner. The PM does not forward the output blindly. The PM verifies the source trail, checks the underlying Planner tasks, confirms whether the evidence is complete, and routes the recommendation to the correct decision owner.”Presenter script
Board update · validation checkpoint active
Cowork produced a risk synthesis in Act 2. Before anything is escalated, the governance checkpoint is now live on the board: the named PM is verifying the synthesis against the underlying tasks. Accountability stays with the human — and the board shows it.
EU AI Act Conformity Readiness Programme
Planner · Board view
Executive Governance
2Prepare executive decision memo
Prepare executive decision memo for governance review
Cannot be finalised until legal and privacy inputs are validated.
Validate Cowork risk scan before executive use
Validation in progress — Diane is checking the underlying Planner tasks and evidence completeness.
Act 4
Source traceability and human review
Recommendations must link back to underlying tasks, files, or evidence — and say plainly when they can't.
- —Not every AI output is automatically executive-ready
- —Some items should be marked "needs additional evidence" — that is a feature of the model, not a weakness
- —The PM role gets stronger: a structured validation queue, not a blind forward
# Objective For the programme risk scan you prepared, create a source validation view. # Instructions For each major risk or recommendation: 1. Identify the related Planner task. 2. Identify the source or evidence used — linked files, task notes, meeting references, comments, or available communications. 3. State whether the evidence is sufficient for PM review. 4. Identify what a human reviewer must validate before this can be used in an executive briefing. 5. Mark each item as: - Ready for PM validation - Needs additional evidence - Requires Legal review - Requires Privacy review - Requires Security review - Requires Product confirmation # Output Format A validation table with: Risk or recommendation / Related Planner task / Source or evidence reference / Evidence sufficiency / Human validation owner / Required next step / Executive briefing readiness. Do not assume evidence exists if it is not available.
Act 5
Cross-functional access boundary scenario
Not every stakeholder needs the same information. Product needs to know a dependency exists — not the privileged legal reasoning behind it.
“Now let us introduce a realistic CELA constraint. Product may need to know that a legal dependency exists. It may not need privileged legal reasoning. Executives need decision implications. Communications needs approved language only.”Presenter script
# Objective Create stakeholder-specific communication views for the EU AI Act Conformity Readiness Programme. # Context The programme has different stakeholder groups with different information needs. # Instructions Using the available Planner and Microsoft 365 context, prepare separate communication drafts for: 1. Executive leadership 2. Legal team 3. Privacy team 4. Product team 5. Security team 6. Communications team For each stakeholder group: - include only the information relevant to that group - avoid unnecessary legal or sensitive detail - distinguish between confirmed facts, pending reviews, and recommendations - identify what action is needed from that group - state where human approval is required before sending # Output (per stakeholder group) ## Audience ## Purpose of message ## Draft message ## Information sensitivity note ## Required human approval before sending Do not send or finalise any communication without human review.
Act 6
Executive decision memo
Executive-ready material — but only after validation logic has already been demonstrated in Acts 3 and 4.
- —Highlight the Ready / At Risk / Not Ready / Awaiting Validation classification
- —Highlight "items not ready for executive escalation" — named explicitly, not buried
- —Highlight named human owners against every open decision
# Role Act as a CELA PM briefing assistant. # Objective Prepare an executive decision memo for the EU AI Act Conformity Readiness Programme. # Important Governance Constraint This memo is a draft for human review. It must not be treated as the final legal, policy, privacy, or executive position. # Inputs to Use Validated Planner task status, confirmed risks, approved evidence references, known approval bottlenecks, documented pending decisions. # Output Structure ## 1. Executive Summary ## 2. Current Readiness Position — classify each workstream: Ready / At Risk / Not Ready / Awaiting Validation Workstreams: Legal, Privacy, Product, Security, Public Policy, Communications, Executive Governance ## 3. Key Risks — description, workstream, decision impact, mitigation, follow-up owner ## 4. Decisions Required — decision, options, recommended path, implication of delay, accountable owner ## 5. Items Not Yet Ready for Executive Decision ## 6. Recommended Governance Actions # Tone Concise, factual, risk-aware. Do not overstate readiness. Do not invent legal conclusions. Do not present recommendations as approved decisions.
Act 7
Coordination in action — calendar, meeting, and team messages
This is where the demo moves from synthesis to real execution. Cowork checks calendar availability, schedules the governance review meeting, and drafts individual Teams messages to task owners — every action still gated by PM approval before it leaves the workspace.
Calendar & meeting
# Objective Check calendar availability for the Legal reviewer, Privacy lead, Security reviewer, and Executive sponsor over the next three business days. # Instructions Propose two meeting time options for a 45-minute "EU AI Act Conformity Readiness — Governance Review" session, based on genuine overlap in their calendars. Do not schedule anything yet — present the two options to the PM for confirmation first. Once the PM confirms a time, schedule the meeting with an agenda covering: - outstanding legal interpretation items - privacy assessment status - security control evidence - governance sign-off recommendation Do not schedule the meeting until the proposed time is explicitly confirmed by the PM.
EU AI Act Conformity Readiness — Governance Review
Scheduled ✓ confirmed via Outlook Calendar[Date] · [Time] · 45 min · Microsoft Teams
Agenda
- —Outstanding legal interpretation items — Legal
- —Privacy assessment status — Privacy
- —Security control evidence — Security
- —Governance sign-off recommendation — Executive sponsor
"Coordination isn't only synthesis — it's action. Cowork checked calendar availability across Legal, Privacy, Security, and our executive sponsor, and found a slot that actually works for all four. I confirmed the time before anything was scheduled. Now it's on every calendar, with an agenda that maps directly to the open items in Planner."Presenter script
Team messages
# Objective Draft individual Teams messages to each Planner task owner with an overdue or at-risk item ahead of the governance review meeting. # Instructions For each task owner: - name the specific task and its current status - state what is needed from them before the review meeting - include the confirmed meeting time - keep the tone direct and collegial Output one message block per recipient, labelled with name and role, followed by "Draft — pending PM approval." Do not send any message until the PM has reviewed and approved the draft.
Legal reviewer
Legal Review
Hi — the legal interpretation note on Article 6 classification is still open in Planner. Could you confirm status before our review at [time] on [date]? Flagging it as the one open item Privacy is waiting on to close their assessment.
Privacy lead
Privacy & Data Protection
Hi — the GDPR impact assessment is still pending your sign-off. Bringing it to the review at [time] on [date] as a decision item — could you have a position ready going in?
Security reviewer
Security Review
Hi — the security control evidence attachment is still open. Could this be closed out before [time] on [date]? It's the last item standing between Security and a Ready status.
Act 8
Human approval before action
Explicitly shows that stakeholder communication or executive escalation requires human review — before anything leaves the PM workspace.
- —PM validates Cowork executive memo
- —Legal confirms legal risk language
- —Privacy confirms privacy assessment status
- —Security confirms control evidence
- —Executive sponsor approves the governance pack
- —Communication sent only after approval
“Before this briefing leaves the PM workspace, the workflow requires human validation. Legal reviews legal language. Privacy validates privacy status. Security confirms evidence. The PM ensures the programme narrative is accurate. Only then does the output become an executive communication or stakeholder action.”Presenter script
Board update · after human approval
Now the board reflects governed execution. The overdue legal note has been validated and signed off, the blocked policy task is released, the unowned regional task has a human owner, and the validation checkpoint is closed — every change made by an accountable person, with Cowork only surfacing where attention was needed.
EU AI Act Conformity Readiness Programme
Planner · Board view
Legal Review
1Complete legal interpretation note
Complete legal interpretation note on high-risk classification
Signed off by Legal — dependency cleared.
Policy & External Affairs
1Draft public policy position
Draft public policy position statement
Unblocked — legal interpretation note now finalised.
Executive Governance
2Prepare executive decision memo
Prepare executive decision memo for governance review
Legal and privacy inputs now validated; memo in executive review.
Validate Cowork risk scan before executive use
Checkpoint closed — human validation complete.
Implementation Tracking
1Confirm regional stakeholder impacts
Confirm regional stakeholder impacts (EU markets)
Owner assigned by the PM after Cowork surfaced the coordination gap.
▸Optional prompt — approval checklist generator
Create a human approval checklist for the executive decision memo. Include: PM validation, Legal review, Privacy review, Security review, Product confirmation, Executive sponsor approval. For each approval step, specify: what must be checked, who should validate it, what evidence should be reviewed, what should happen if the reviewer disagrees, how the Planner task should be updated after approval.
Act 9
Outcome and measurement layer
A CELA PM team will not adopt this workflow because it looked impressive. It has to move real programme metrics.
| Metric | Before | After | Target direction |
|---|---|---|---|
| Risk identification cycle time | Baseline | Measured | Reduce delay |
| Executive briefing preparation effort | Baseline | Measured | Reduce manual effort |
| Approval bottleneck visibility | Baseline | Measured | Improve earlier escalation |
| Stakeholder follow-up completion | Baseline | Measured | Improve closure rate |
| Decision ageing | Baseline | Measured | Reduce unresolved decisions |
| Evidence completeness | Baseline | Measured | Improve readiness quality |
Replace "Baseline / Measured" with real pilot figures before delivery — an unquantified table invites the exact challenge it is meant to pre-empt.
▸Measurement framework prompt
# Objective Create a measurement model for the Planner + Cowork CELA PM workflow. # Context The workflow supports an EU AI Act Conformity Readiness Programme involving Legal, Privacy, Product, Security, Policy, Communications, and Executive Governance. # Output Create a measurement framework with: Adoption metrics, Governance metrics, Coordination metrics, Risk management metrics, Executive decision metrics, Evidence quality metrics. For each metric: name, what it measures, why it matters, possible data source, suggested review cadence, owner accountable for reviewing it. Avoid claiming improvement without measured evidence.
Closing script
“The real value of Planner and Cowork for CELA PMs is not faster task management. It is trusted coordination. Planner holds the commitments. Cowork accelerates analysis, synthesis, communication drafting, and decision preparation. Human owners validate the evidence, approve the recommendations, and remain accountable for decisions. For CELA, that is the right model: faster coordination without weakening governance.”
Leave-behind: the master briefing prompt
The full end-to-end prompt a CELA PM can keep and reuse. Run it live to generate a complete governed programme briefing from the board above.
# Role Act as a CELA Programme Management assistant supporting a legal, privacy, policy, product, security, communications, and executive governance programme. # Scenario We are managing an EU AI Act Conformity Readiness Programme in Planner. Planner is the system of record for programme commitments, including tasks, owners, due dates, workstreams, dependencies, evidence links, and approval actions. # Objective Prepare an executive-ready programme briefing that supports governance review and decision-making. # Governance Constraint This output is a draft for human review. Do not present legal, privacy, security, regulatory, or executive conclusions as final unless explicitly confirmed in the available source context. Clearly distinguish between: confirmed facts, pending reviews, recommendations, assumptions, information gaps. Do not invent owners, dates, approvals, evidence, or legal conclusions. # Analyse Review available Planner tasks and related Microsoft 365 context where accessible. Identify: overall programme health, workstreams on track / at risk / blocked, overdue tasks, missing evidence, critical path dependencies, pending approvals, stakeholder groups requiring follow-up, decisions required from leadership, risks with legal, privacy, security, regulatory, reputational, policy, or delivery implications. # Structure ## 1. Executive Summary ## 2. Programme Health — On Track / At Risk / Off Track / Awaiting Validation, with evidence-based rationale ## 3. Workstream Status — Regulatory Interpretation, Legal Review, Privacy & Data Protection, Product Readiness, Security Review, Policy & External Affairs, Communications Readiness, Executive Governance, Evidence & Audit Trail ## 4. Key Risks and Implications — prioritised, with severity and validation owner ## 5. Decision Requirements — for senior leadership, with urgency and required validation ## 6. Source and Evidence Validation — classify each conclusion: Ready for executive use / Needs PM validation / Requires Legal, Privacy, Security, or Product review / Not enough information available ## 7. Stakeholder Coordination Plan — Legal, Privacy, Product, Security, Public Policy, Communications, Executive Sponsors, Regional or Market Teams ## 8. Immediate PM Actions ## 9. Executive Talking Points — 5 to 7, clear, factual, risk-aware, non-alarmist, action-oriented ## 10. Draft Communications — executive briefing email, governance meeting agenda, Legal/Privacy follow-up note, risk escalation note, stakeholder coordination message. Each draft states the human approval required before sending. # Output Requirements Polished executive format. Avoid generic project management language. Focus on governance discipline, legal and regulatory readiness, privacy and security validation, stakeholder alignment, decision velocity, evidence completeness, risk mitigation, execution confidence. Where information is missing, state: "Information not found or not available in the current context."