Skip to content

AD 102 · Module 1

Interview and Current State

Learn structured interview principles, ask about the work itself rather than AI, map the current-state process, and mark handoffs and decision points.

Start Section 1
0 / 5 completed
Sync my progressJoin a session to keep progress across devices and share it with your instructor

Info

After you join, your completed lesson numbers are sent to your instructor together with your learner code, so they can see your progress. Use the codes your instructor gave you, and do not enter your name, company secrets or other personal data. Progress is kept with the rest of that session's data.

Learning objectives

  • Use neutral, concrete-event, evidence-probing questions and avoid leading questions.
  • Ask about the work through trigger, input, actor, tool, decision, and exception, never start with "what would you like AI to do".
  • Build a bounded current-state sequence that preserves actor, system, input, and output.
  • Locate handoffs, queues, approvals, duplicate entry, and key decisions.
  • Conduct a paired or small-group interview, map while listening, and finish with a member check.

Discovery Interview Principles

30 min

Discovery Interview Principles

Learning objective: Use neutral, concrete-event, evidence-probing questions and avoid leading questions.

A leading question already implies the answer, e.g. "isn't this process really inefficient?", the interviewee tends to agree without describing facts. A neutral question asks about a concrete event instead: "what happened the last time you handled this?" Evidence probing follows an impression-based answer with a request for specifics: "you said 'often', roughly how often? When was the last time?"

Interview-question self-check
FieldMeaning
Does it hint at the answer?Yes -> rewrite as an open question
Is it about an impression or an event?Impression -> ask for the specific last time
Does it ask for evidence?No -> add "how do you know that"
Workplace example

Interviewing an admin office about its leave-request process. Leading: "isn't sign-off really slow?" Neutral + evidence probing: "the last time you submitted a leave request, how long did it take your manager to approve it? If you remember, what date did you submit it and what date was it approved?"

Leading question (hints at the answer)

  • "Isn't this system really hard to use?"
  • "Doesn't everyone think this process is too complicated?"
  • "Isn't approval always really slow?"

Neutral + evidence probing (asks for the event)

  • "The last time you used this system to finish a task, how long did it take, and where did it get stuck?"
  • "What actually happened the last time you went through this process?"
  • "Do you remember the last approval: what date was it submitted, and what date was it approved?"
  1. Ask an open question
    Do not presuppose a direction; ask "what happened" rather than "isn't this bad".
  2. Pause on impression words
    When someone says "often", "a long time", or "everyone", note it down but do not treat it as fact yet.
  3. Probe for the specific last time
    "When was the time you're thinking of? How long did that one take?" This pulls an impression back to a concrete event.
  4. Probe for the source of evidence
    "How do you know that? Did you experience it yourself, or hear it from someone else?" This checks how reliable the information is.

Info

Human, AI, Human: a person leads the interview and judgment; AI may help organize notes, but the process conclusions and evidence judgment stay with the person.

Warning

Data safety reminder: every exercise in this course uses course-provided synthetic scenarios only, never real institutional or personal data.
Check yourself

Which question best fits the neutral, concrete-event, evidence-probing principle?

Which question best fits the neutral, concrete-event, evidence-probing principle?

Key takeaway: A good interview question does not hint at the answer, it draws out a specific past event and asks for evidence.

Ask About Work, Not AI

35 min

Ask About Work, Not AI

Learning objective: Ask about the work through trigger, input, actor, tool, decision, and exception, never start with "what would you like AI to do".

Opening with "what would you like AI to do" tends to make the interviewee imagine a feature rather than describe real work. Ask instead about six facets of the work itself: what triggers it, what inputs come in, who is involved, what tools are used today, what decisions occur, and what exceptions arise. Only once the work itself is clear can you judge whether AI could help.

Six facets of asking about the work
FieldMeaning
TriggerWhat starts this work?
InputWhat data or documents come in?
ActorWho is involved?
ToolWhat system or tool is used today?
DecisionWhat judgments occur along the way?
ExceptionWhat exceptions or special cases exist?
Workplace example

Interviewing the leave-request process, instead of "what would you like AI to do", ask: "Trigger: what does an employee open when they want time off? Input: which fields do they fill in? Actors: who sees the request? Tool: what system or form is used today? Decision: what does the manager weigh when approving? Exception: is there ever a case that skips the normal flow?"

Workplace example (contrast case: venue-booking request)

Interviewing a venue-booking process, the same six facets apply: Trigger (a club checks venue availability when planning an event), Input (event name, headcount, time slot), Actor (the requesting club, the venue office, the fire-safety reviewer), Tool (currently a paper form plus an Excel schedule), Decision (whether headcount exceeds capacity, whether extra security is needed), Exception (large events need two weeks' notice, shared venues need both sides' agreement). Clarifying these six first is what lets you judge whether, and where, AI could actually help.

  • Opening with "what feature do you want" tends to make the interviewee picture an interface or chat window, skipping the real work detail.
  • Breaking the question into six facets helps avoid missing exceptions, which are the facet most often overlooked.
  • Ask the Tool question about what is used "now", not what people wish they had, so you learn the actual current state.
  • Push the Decision facet to "what basis is the judgment made on", not just the surface answer "a manager reviews it".

Info

Human, AI, Human: a person leads the interview and judgment; AI may help organize notes, but the process conclusions and evidence judgment stay with the person.
Check yourself

At the start of an interview, which opening question best fits this lesson's principle?

At the start of an interview, which opening question best fits this lesson's principle?

Key takeaway: Clarify the work's trigger, input, actor, tool, decision, and exception first, AI's role comes later.

Map Current Process

40 min

Map Current Process

Learning objective: Build a bounded current-state sequence that preserves actor, system, input, and output.

A current-state process map arranges what you learned in the interview into a bounded sequence of steps, each labeled with who does it (actor), which system is used (system), what comes in (input), and what goes out (output). The scope must be bounded: a clear start and end, not the whole department's work crammed into one diagram.

Four labels per step
FieldMeaning
ActorWho performs this step?
SystemWhat system or tool is used?
InputWhat comes into this step?
OutputWhat comes out of this step?
Workplace example

Leave-request current-state map: (1) employee fills out the form (actor: employee; system: paper form; input: reason for leave; output: completed form) (2) manager reviews (actor: manager; system: none; input: completed form; output: approval or rejection) (3) HR logs it (actor: HR; system: HR system; input: approved form; output: leave record).

  1. Set the boundary first
    Write down exactly where step one starts and where the last step ends before worrying about whether the map will be complete.
  2. List steps in time order
    Write down each step in the order it actually happens, without trying to simplify or clean it up yet.
  3. Add the four labels to each step
    Label actor, system, input, and output for every step; leave a label blank and flagged if it is not yet confirmed.
  4. Check that steps connect
    Confirm each step's output matches the next step's input; a gap here is often a missed step.
Workplace example (contrast case: venue-booking request)

Venue-booking current-state map: (1) club fills out an application (actor: club representative; system: paper form; input: event details; output: completed application) (2) venue office checks availability (actor: venue staff; system: Excel schedule; input: application; output: a yes/no on availability) (3) fire-safety review (actor: security team; system: none; input: headcount and venue details; output: whether extra security is needed) (4) approval notice (actor: venue staff; system: email; input: review result; output: approval or rejection notice).

Info

Human, AI, Human: a person leads the interview and judgment; AI may help organize notes, but the process conclusions and evidence judgment stay with the person.
Check yourself

When drawing a current-state process map, which approach is most correct?

When drawing a current-state process map, which approach is most correct?

Key takeaway: A current-state map needs a clear boundary, with every step labeled by who, what system, and its input/output.

Decisions and Handoffs

35 min

Decisions and Handoffs

Learning objective: Locate handoffs, queues, approvals, duplicate entry, and key decisions.

A handoff is where work moves from one person or system to another, often a source of delay and lost information. A queue is where work waits to be processed. Approvals and key decisions need their basis and owner recorded. Duplicate entry is the same data being entered more than once, a common source of waste and error.

Four signals
FieldMeaning
HandoffWho does the work move from and to?
QueueWhere does work wait?
Approval / decisionWho approves, and on what basis?
Duplicate entryIs the same data entered more than once?
Workplace example

In the leave process: the employee hands the paper form to the manager (handoff 1, possible loss or delay); after approval, HR re-enters the same data into the system (handoff 2 + duplicate entry, since HR retypes what the employee already wrote).

A healthy handoff

  • There is a clear receipt confirmation when work changes hands (a system notice, a signature).
  • Waiting has a cap, and someone actively follows up if it is exceeded.
  • Approval criteria are spelled out, so different reviewers reach consistent results.

A problem handoff

  • Once a case is passed on, no one knows if it was received; you can only wait for the other side to "remember".
  • The queue has no cap; a case can sit untouched for days.
  • The same data is retyped into a different system, with no one checking the two match.
  • Ask: "Once it's handed off, how do you know the other side received it?" No clear answer means a potential handoff risk.
  • Ask: "How long does a case usually sit here?" An answer like "it depends how busy they are" means an uncapped queue.
  • Ask: "What conditions matter most when approving?" Different answers from different managers mean inconsistent approval criteria.
  • Ask: "Is this data also typed in somewhere else?" This quickly surfaces duplicate-entry points.

Info

Human, AI, Human: a person leads the interview and judgment; AI may help organize notes, but the process conclusions and evidence judgment stay with the person.
Check yourself

You notice HR retypes the leave reason the employee already wrote into the system. Which signal is this?

You notice HR retypes the leave reason the employee already wrote into the system. Which signal is this?

Key takeaway: Handoffs, queues, approvals, and duplicate entry are the four signals most worth marking on a process map, and often the starting point for pain-point analysis.

Guided Interview Lab

40 min

Guided Interview Lab

Learning objective: Conduct a paired or small-group interview, map while listening, and finish with a member check.

Work in pairs: one interviews, one records and sketches the process live. After the interview, do a member check: read the drawn map back to the interviewee and ask, "did I get this right? Did I miss or misdraw anything?" The member check is the key step that stops the interviewer from filling gaps with their own assumptions.

Four steps of the guided lab
FieldMeaning
Split rolesOne interviews, one records and sketches
InterviewUse neutral questions, probe for evidence
SketchDraw the current-state process while listening
Member-checkRead it back to the interviewee to confirm accuracy
Workplace example

Scenario: one classmate plays an "administrative staff member" describing a real or simulated task (e.g. a venue-booking request). Another interviews and sketches the process live, then reads the map back to the interviewee for step-by-step confirmation.

  1. Split roles and open the conversation
    Agree who asks and who records and sketches, and explain the purpose up front to lower the interviewee's guard.
  2. Sketch while interviewing
    The recorder draws step boxes as they listen; do not worry about neatness yet, the priority is not missing a step.
  3. Probe the gaps on the spot
    When a drawn step is missing an actor or input, ask right then, do not wait until the interview ends to notice.
  4. Close with a member check
    Read the whole map back and ask step by step, "is this right? Any case I missed?", then revise based on the answer.
  • "Did this step happen before or after that one?" confirms the order is right.
  • "Is there a step here I haven't drawn yet?" actively invites the interviewee to fill a gap.
  • "Is this decision yours to make, or do you need to check with someone?" confirms the actor and decision are drawn correctly.
  • "Would the flow look different for a special case?" confirms exceptions were captured.

Info

Human, AI, Human: a person leads the interview and judgment; AI may help organize notes, but the process conclusions and evidence judgment stay with the person.
Check yourself

After the interview, which step best prevents the interviewer from filling in gaps with assumptions?

After the interview, which step best prevents the interviewer from filling in gaps with assumptions?

Key takeaway: Sketch while listening, and always finish with a member check, the interviewee confirms the map, not the interviewer's own assumptions.

What next?