Practical guide
AI fluency for programme managers: track a shared benefit
Map a fictional programme benefit across dependent workstreams without adding their overlapping outcome estimates.
Programme managers can use AI to organize cross-workstream dependencies while checking how a benefit would arise. This fictional service-improvement programme combines a new intake form and routing changes. It produces a benefit map and evidence questions, not a promised return on investment or a sum of independently invented savings.
Name the benefit separately from the deliverables
A completed component is not the same as the outcome the programme hopes to achieve.
The fictional programme has two workstreams: a clearer request form and automatic routing to the appropriate team. Both aim to reduce requests being sent back for clarification. The form and routing configuration are deliverables; fewer clarification loops is the proposed benefit. The packet supplies no measured improvement.
Ask AI to draw the causal proposal in words: what each component changes, what user behavior must follow and what observable outcome would matter. Keep these links as hypotheses until evidence exists. Do not mark the benefit realized because both workstream status reports show delivery complete.
Identify where benefit estimates overlap
Two estimates can describe the same avoided work rather than two separate gains.
Suppose each workstream’s planning note claims ten avoided clarification loops from the same set of twenty requests. Adding the claims to twenty avoided loops could double count the same cases. The packet does not identify which requests each workstream would affect or how the changes interact.
Create a shared outcome register with the request population, benefit definition and attribution uncertainty. Ask AI to find overlap in the proposed mechanism, not merely duplicate wording. Keep component estimates separate from a programme total until the relationship between them is understood and the evidence supports aggregation.
Map the prerequisites for the benefit
A component can be available while a necessary operating condition is still missing.
The revised form must be used, requesters must understand its fields and the receiving teams must recognize the routing categories. The fictional packet says the form is deployed but an old bookmarked link remains in circulation. That means deployment alone does not establish that incoming requests use the new form.
Record an owner and evidence question for each prerequisite. A page-view count might indicate exposure but not whether the submitted information is complete. Ask which observation would distinguish the proposed benefit mechanism from mere tool usage. Do not replace outcome evidence with a convenient activity metric.
Handle a partial rollout without rewriting the objective
When one workstream slips, show what can still be learned and what cannot.
Introduce a delay in the routing change while the new form remains available. The programme might inspect form completion issues, but it cannot attribute a combined end-to-end outcome to the full programme before both components and their prerequisites exist. Keep the observation scope explicit.
Do not ask the form team to absorb an unapproved routing responsibility simply to maintain a green programme summary. Present the dependency, possible sequencing choices and decision owner. AI can help explain alternatives, but it cannot approve extra scope or declare a benefit achieved without supporting observations.
Deliver a benefit map with an evidence gate
A programme update should distinguish delivery, use and outcomes.
Include the two workstreams, shared request population, overlapping estimates, prerequisites and current rollout state. Define the next evidence needed to judge whether clarification loops changed and what other factors could explain a difference. Do not manufacture a baseline or percentage saving to complete the report.
Review whether the worker catches double counting and preserves the old-link adoption problem. The project manager exercise focuses on task sequencing and capacity. This programme guide focuses on how several delivered changes might jointly produce one benefit and why their estimates cannot simply be added.
Sources and scope
NIST: AI risk management. Skills England: workplace AI foundations.
These references provide background, not validation or endorsement of this exercise. The case details, calculations and suggested review questions are original instructional material. Use them to discuss observable work, not to infer customer outcomes, professional credentials or performance in every setting. Before adapting the exercise, confirm the relevant facts, approved tools, data permissions and decision owners. If you change the case, revisit the expected answers and checks as well. These examples describe practice tasks, not a promise that a particular product includes the fictional features.
Sources: [1] [2]
Sources
- 1.AI RMF Core · NIST
- 2.AI foundation skills for work benchmark · Skills England