Practical guide
Communication and delivery: make the next action clear
Adapt a fictional service-window change for executives and frontline managers while preserving facts, conditions and ownership.
Communication and delivery means turning completed analysis into an artifact the intended reader can use without guessing what is approved, uncertain or required next. This original service-window exercise separates communication quality from visual polish. It shows how to preserve the same factual core across different audiences while changing structure, level of detail and the action the document supports.
Name the action the reader needs to take
Choose the document structure after identifying its audience and decision.
In the fictional exercise, a support window may move by one hour. Executives need to approve a proposal; frontline managers need instructions only after that approval. A single enthusiastic announcement cannot serve both purposes safely. Write the intended action at the top of each brief before drafting the rest.
For the executive version, lead with the requested decision and its conditions. For the manager version, lead with the operational change, effective date and required preparation. If approval is still pending, keep the manager document clearly marked as a draft. Tone should not silently convert a proposal into an instruction.
Preserve the factual core across versions
Adapting a message means selecting and arranging information, not changing its status.
Create a compact fact sheet with the proposed hours, affected teams, unresolved staffing question and evidence used to assess customer impact. Use it to compare the two drafts. If one says the change is approved while the other says it is proposed, resolve the discrepancy before distribution.
AI can help produce different formats, but inspect dates, quantities, scope and approval language after transformation. A shorter draft can accidentally remove a condition that changes the meaning. Preserve necessary uncertainty even when it makes the sentence less tidy or prevents a punchier headline.
Make evidence proportionate to the claim
Choose wording that reflects what the supplied material can actually establish.
The fictional pack includes a small set of customer comments. Those comments can illustrate concerns; they do not establish that all customers prefer the new window. Describe the source and limitation rather than turning selected feedback into a universal statement. If broader evidence is needed for approval, identify that gap explicitly.
Place the supporting reference close to the decision it informs. Avoid sending a long appendix without explaining why it matters. A reader should be able to distinguish observed feedback, the author's interpretation and the proposed response. This separation is more useful than filling the memo with confident but untraceable language.
Build an operational handoff
Specify who must do what, by when, and what dependency could prevent completion.
The manager checklist might include confirming roster coverage, updating the team notice and reporting unresolved customer exceptions. Each item needs an owner and a timing condition tied to approval. Do not use a date invented by the assistant just because a complete checklist looks more professional.
Ask a cold reader to identify the first action they would take. If they cannot distinguish preparation from launch, revise the headings or sequence. Test whether they can find the escalation route for a staffing problem without searching the entire document. This is a usability check within the exercise, not a survey of communication skill.
Release a coherent final artifact
Before sending, check that the final version is the one the reader is meant to act on.
Compare the executive memo, manager checklist and fact sheet for consistency. Remove obsolete draft attachments, preserve a version label and make the approval status visible. Check links and attachments as part of delivery; a clear sentence cannot compensate for a missing schedule or an inaccessible supporting file.
To finish the exercise, introduce a delayed approval and ask the author to update both outputs. Look for changes to timing, status and owner instructions while unchanged facts remain stable. Judge the work by whether the intended reader can act appropriately on these materials, not by decorative formatting or a fixed preference for short documents.
Sources and scope
NIST addresses AI risk management. Skills England describes workplace AI foundations.
These sources provide background, not endorsement of this exercise. The worked example and suggested review method are original illustrative guidance. They are not customer results, validated benchmarks or evidence of a particular product capability. Adapt the exercise to the task and use qualified review where consequences require it.
Sources: [1] [2]
Sources
- 1.AI RMF Core · NIST
- 2.AI foundation skills for work benchmark · Skills England
Explore this topic
- Action orientation in AI-enabled work: make the handoff executable
Action orientation is the ability to turn analysis into a clear, evidence-linked handoff without hiding uncertainty or taking authority that belongs to someone else. A useful deliverable tells the reader what decision is needed, who owns the next step, what evidence supports it, and what should happen if a condition changes.
- Audience adaptation in AI-enabled work: preserve the decision, change the delivery
Audience adaptation is the ability to reshape a work product for a specific reader without changing the underlying evidence or hiding important uncertainty. The objective is not to make the same message sound friendlier. It is to help each audience make the decision, ask the question, or complete the action that belongs to them.
- Claim calibration in AI-enabled work: match language to evidence
Claim calibration is the ability to express a conclusion at the strength the available evidence supports. It distinguishes what was observed from what was inferred, estimated, proposed, or left unresolved. In AI-enabled work, calibration is essential because fluent drafting can make weak support sound settled.
- Delivery completeness in AI-enabled work: close the decision, not just the document
Delivery completeness is the ability to produce a work artifact that contains everything its reader needs to make the intended decision or carry out the intended action. A polished draft can still be incomplete when it omits a dependency, source, owner, limitation, approval, or next step. Completeness is therefore judged against the assignment and handoff, not against length.
- Revision quality in AI-enabled work: improve the artifact without losing its evidence
Revision quality is the ability to turn feedback, new evidence, or a failed test into a better work product while preserving traceability. It is not the number of edits made, the smoothness of the prose, or the willingness to accept every suggestion. A strong revision fixes the diagnosed problem, protects sound material, and leaves a reviewer able to see what changed and why.
- How to develop and assess source citation in AI-enabled work
Source citation is the ability to show where a consequential statement came from, which version of the source was used, and how closely the statement matches the underlying evidence. In AI-enabled work, this is more than adding a list of links after drafting. The useful behavior happens while claims are selected, checked, narrowed, and connected to sources a reader can inspect.
- How to develop and assess structure and hierarchy in AI-enabled work
Structure and hierarchy turn a collection of correct points into an artifact a reader can use. The skill is not simply writing headings. It is deciding what the reader must understand first, grouping evidence around that decision, and making supporting detail available without allowing it to obscure the main message.
- How to develop and assess visual explanation in AI-enabled work
Visual explanation is the ability to choose and build a visual that helps a specific reader understand evidence, comparison, sequence, or causation without distorting it. In AI-enabled work, the skill includes rejecting unnecessary graphics, checking the underlying data, naming the intended takeaway, and providing an equivalent explanation for readers who cannot use the image.