AI-Assisted Requirements Gathering and Stakeholder Analysis
Deliberate Academy Editorial Team
Reviewed for accuracy and professional relevance
Enjoying the course?
Sign up free to track your progress and earn a verified certificate when you pass.
- Use AI to generate a structured requirements workshop agenda and interview question framework from a project description and stakeholder group list
- Apply the five-area review checklist to AI-generated user stories and acceptance criteria to surface the gaps AI consistently underrepresents
- Use AI conflict identification to surface logical inconsistencies in a requirements set that self-review might overlook
- Explain why AI-generated stakeholder analysis documents require substantial revision from someone with organizational knowledge
Requirements gathering is the activity that most determines whether a project delivers something useful. It is also the activity where the gap between what was discussed and what was documented is widest, and where AI can both help and mislead in equal measure. Used well, AI compresses the administrative burden of requirements work and improves the structure of the artifacts produced. Used carelessly, it generates requirements that sound complete while missing the substance that only stakeholder conversation surfaces.
Preparing for Requirements Workshops with AI
The work that happens before a requirements workshop largely determines its quality. A BA who arrives with well-framed questions, a clear stakeholder map, and a preliminary view of the risk areas to probe will extract more useful information than one who arrives with a blank document and good intentions.
Generating interview question frameworks. Given a project description, the affected business processes, and the stakeholder groups involved, AI can generate a structured set of workshop questions — organised by stakeholder type, process area, and risk category — in a fraction of the time it would take to produce manually. These should be treated as a starting set to be reviewed and refined, not a final agenda. AI does not know which questions are politically sensitive in your organization, which stakeholders have competing interests, or what was promised in a previous project that failed.
Drafting preliminary persona hypotheses. If you have initial information about the user groups the system or change will affect, AI can generate persona drafts that describe their likely goals, pain points, and success criteria. These are useful as pre-read material for workshop participants and as a structure to confirm or challenge in conversation. They are not substitutes for the conversation — personas created without user input are assumptions, not findings.
Identifying risk prompts. A prompt structured around the project type, the affected systems, and the stakeholder groups will typically generate a useful list of risk areas to probe — regulatory dependencies, integration points, user adoption challenges, data quality concerns — that an experienced BA might develop independently but a less experienced one might miss entirely.
AI-Assisted User Story Generation
User stories are where AI is most widely used in BA work and where the quality risk is most frequently underestimated. AI can convert rough notes into user story format quickly and consistently. The result looks professional. The question is whether it is accurate and complete.
From rough notes to structured stories. A well-constructed prompt — providing the project context, the user group, the business process being changed, and the notes from a discovery conversation — can yield a draft set of user stories formatted correctly and covering the obvious journeys. See Prompting AI Effectively for the general structure this applies. This is a legitimate time-saver. The draft requires careful review against the source material and against the stakeholder who provided the notes.
Acceptance criteria drafting. AI can produce acceptance criteria for a given user story efficiently. The output typically covers the obvious success conditions. The non-obvious ones — the edge cases, the error paths, the performance conditions, the accessibility requirements, the audit trail requirements — are far less reliably present and must be systematically added through BA review and stakeholder confirmation.
Requirements conflict identification. Providing AI with a set of user stories or a requirements document and asking it to identify apparent conflicts or ambiguities is a useful quality check. AI will surface logical inconsistencies — two requirements that cannot both be true, or a requirement that contradicts a stated constraint — that a BA reviewing their own work might overlook. This is one of the more reliable AI quality applications in requirements work.
Use a structured review checklist when reviewing AI-generated user stories and acceptance criteria. The checklist should systematically probe five areas that AI consistently underrepresents: non-functional requirements (performance, security, accessibility, scalability), error and exception paths, edge cases at the boundary of the described scope, audit and compliance requirements specific to your industry, and any requirement that depends on organizational context the AI was not given. A brief, consistent checklist used on every AI-generated artifact catches far more gaps than an ad hoc review.
Cutting requirements workshop preparation time at a retail bank
Context
A senior BA at a retail bank was leading requirements discovery for a new mortgage origination system — a complex project with twelve stakeholder groups across front office, compliance, credit risk, and IT. Workshop preparation — developing the question framework, preliminary persona drafts, and a risk prompt list — typically took two to three days and was the step most likely to be compressed when project schedules tightened.
Action
The BA used AI to generate a structured question framework across all twelve stakeholder groups from the project description and business process overview. She also generated preliminary persona drafts for the five primary user types. After each workshop, AI-generated user stories were reviewed using the five-area checklist — specifically checking for non-functional requirements, error and exception paths, and compliance requirements specific to mortgage regulation. Two rounds of checklist review surfaced nine acceptance criteria gaps before the requirements baseline was agreed.
Outcome
Workshop preparation time fell from two to three days to under one day without reducing the depth of coverage in the sessions. The checklist review process caught requirements gaps that would previously have been identified by the development team during sprint review — at significantly higher cost to fix. The compliance team noted that the AI-generated question framework prompted discussion of several regulatory constraints that had been omitted from the project brief.
A BA uses AI to generate user stories and acceptance criteria for a financial services onboarding system, then reviews only the happy-path journeys before baselining the requirements. Two months into development the team discovers there are no acceptance criteria covering what happens when identity verification fails. Which phase of the review checklist did the BA most critically skip?
Select one answer.
Stakeholder Analysis Documents
Stakeholder analysis — identifying who is affected by a change, what their interests are, how much influence they have, and how they are likely to respond — is one of the most contextually dependent activities in BA work. AI can provide structural scaffolding; the content must come from someone who knows the organization.
Stakeholder map generation. Given a project description and a list of business functions involved, AI can generate a first-draft stakeholder map covering the obvious groups and their likely interests. This is useful as a starting structure. The BA must add, remove, and substantially revise based on organizational knowledge that the AI does not have.
Stakeholder communication planning. AI can generate draft communication plans for different stakeholder groups, adapting tone and content level based on the stakeholder description provided. These are solid first drafts that require review for political appropriateness — AI has no awareness of the interpersonal dynamics between stakeholders or the history of previous change programs.
AI-generated user stories and acceptance criteria sound complete but often miss edge cases, non-functional requirements, and the organizational context that only stakeholder conversation surfaces. A requirements baseline built predominantly from AI-generated stories without robust stakeholder validation creates the most dangerous type of requirements gap — one that is discovered late in development when the cost of change is highest. Use AI to draft, not to replace the conversation, and build in explicit validation steps before AI-generated requirements are baselined.
A BA uses AI to generate acceptance criteria for a set of user stories covering a new customer onboarding process. Which category of acceptance criteria is AI most likely to omit without explicit prompting?
Select one answer.
Exercise
Your Task
Take a set of user stories from a current or recent project — at least five to ten stories with acceptance criteria. Paste them into an AI tool and ask it to identify apparent conflicts or ambiguities between the stories. Then apply the five-area checklist from this lesson to two of the stories yourself: check for non-functional requirements, error and exception paths, edge cases at the boundary of scope, audit and compliance requirements, and organization-specific constraints. Note how many gaps the AI conflict check surfaced that you had not already identified, and how many the checklist review added on top of that.
Your reflection
Did you complete this exercise? What did you find? (Saved locally in your browser)
Try It: AI-Graded Practice
The exercise above is self-assessed. The exercise below is graded automatically, so you can get direct feedback on whether the acceptance criteria you write actually close the gaps this lesson's checklist targets.
- AI is most valuable in requirements preparation — generating workshop question frameworks, preliminary persona drafts, and risk prompts — reducing administrative burden while the BA focuses on the stakeholder conversation that makes those artifacts meaningful.
- AI-generated user stories require systematic review using a checklist that specifically targets non-functional requirements, error paths, edge cases, audit requirements, and organization-specific constraints — the areas AI consistently underrepresents.
- AI can perform useful requirements conflict identification by surfacing logical inconsistencies in a requirements set that a BA reviewing their own work might overlook.
- AI-generated stakeholder analysis documents provide structural scaffolding that requires substantial revision by someone with organizational knowledge — AI has no awareness of interpersonal dynamics, political sensitivities, or the history of previous change programs.
- The most dangerous requirements gap is one discovered late in development — using AI to draft and stakeholders to validate, with explicit baselining controls, prevents AI-generated incompleteness from surviving into delivery.