Practical guide
AI fluency for solutions consultants: bound a demonstration
Build a fictional demonstration plan that separates available behavior, configured examples and unresolved customer requirements.
Solutions consultants can use AI to draft a demonstration while keeping every scenario inside approved evidence. This fictional exercise maps three customer questions to a safe demonstration plan. It produces a scope-and-proof sheet, not a product promise, a contractual commitment or evidence that Two Prune has the described features.
Map questions to demonstration evidence
A demonstration should answer a defined question rather than show every available screen.
The fictional buyer asks whether a report can be exported, whether access can be limited and whether a custom field appears. The approved packet describes manual export and role-based access in a sandbox. It says nothing about the custom field. These are exercise facts only.
Create one row per question with approved source, proposed demonstration step and unresolved gap. AI may draft a narrative, but it must not invent the custom field or imply it can be configured later. A polished sequence is not evidence for an absent capability.
Distinguish a demonstration setup from standard availability
A prepared example may depend on conditions that are not generally present.
The sandbox contains two fictional roles already configured by the demonstration team. State that setup before showing the access difference. Do not imply every account has the same roles or that a demonstration setting defines a production entitlement.
Ask AI to write the audience-facing explanation and an internal prerequisite list. Compare them for consistency. Necessary qualifications should remain close to the claim, while internal credentials or sensitive configuration details stay out of the public narrative.
Design a negative case
Showing what should not happen can expose an important boundary.
For the access question, include an authorized role that can open the report and a restricted role that receives the supplied access message. Record the expected evidence for both. Do not bypass controls or reuse another person’s identity to create the negative case.
This is a paper plan until the sandbox steps are actually run. Keep expected and observed results in separate fields. If the restricted role unexpectedly opens the report, stop the demonstration and route the issue rather than explaining it away in the script.
Handle the unsupported requirement honestly
A gap should produce a clarification path, not a speculative promise.
For the custom field, ask which workflow and decision require it. Offer to confirm whether an approved option exists after the appropriate owner reviews the requirement. Do not state that it is on a roadmap or easy to build without evidence.
Introduce a note from an engineer saying the idea seems feasible. Treat it as a lead for internal review, not approved availability or timing. AI should preserve the source and status so feasibility language does not become a customer commitment.
Deliver a demo runbook with stop conditions
The handoff should show what can be demonstrated, what must be qualified and when to pause.
Include questions, approved evidence, setup, positive and negative cases, custom-field gap, owners and the version of the source packet. Add a stop condition for unexpected access or content. Mark any steps not executed.
Review source fidelity, authority and recovery behavior in this fictional task. Proposal management maps requirements into response text; this guide specifically controls a live demonstration narrative and the evidence visible during it.
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