Sprint One: Discover and Diagnose

Document Collection Checklist

Collect these before discovery begins

Executive Team Survey

Open-ended ethnographic survey · 20-25 min per respondent · 9 core questions + 1 optional · all questions audited against Portigal taxonomy

Audience: Septapod executive team, ~7-8 senior leaders. The CEO takes the same instrument. Optionally extends to the board (Brent's call).

Format note: The platform (Google Form, Tally, etc.) is chosen at delivery time. The questions and order below are the working instrument.

Introduction (framing text, included with the survey)

Brent is preparing for the executive workshop and the per-function task mapping work in Sprint One. These questions are designed to surface how AI is actually showing up in your work today, where you see opportunities and concerns specific to your area, and how the values-based banking framework shapes your thinking. There are no right or wrong answers, and the questions intentionally ask for specific moments and examples rather than ratings. Your responses are seen by Brent and the executive team only, and they shape the Sprint One diagnosis and the workshop's strategic conversation. Estimated time: 20-25 minutes.

Section A: Your work and AI today

Q1. Walk us through a recent moment when AI showed up in your work, sanctioned or otherwise. What were you doing, and what happened? [sequence] Q2. Tell us about a moment in the past six months when AI in a vendor product or tool caught your attention in a way that felt important to flag. What was it, and what did you do? [specific example] Q3. Think of a specific task or process in your area where you or your team spend meaningful time today. What does that task look like end to end? [sequence]

Section B: AI's footprint and trajectory in your area

Q4. Has anything changed in how your team works in the past 12 months because of AI (a vendor feature, a tool someone tried, a process shift)? If yes, what changed? If no, what would have to change for AI to start showing up in your team's work? [specific example + exception] Q5. Name two or three tasks in your area where AI is already showing up or might show up in the next year. For each, briefly describe whether you think the right relationship is: (a) AI runs it with humans on exception review, (b) AI assists a human who makes the call, or (c) it should stay fully human with AI deliberately out of the path. If none of (a)/(b)/(c) feels right for a task, describe the relationship in your own words. This is starting material; do not worry about getting it right. [open-framework] Q6. When a decision about AI comes up that touches more than one department or function, how is it being handled today? Tell us about a recent example if one comes to mind. [relationships + specific example]

Optional closing question

Q7. Eighteen months from now, if Septapod's AI work has gone well, what is something you would notice in your area that is not happening today? [projection, anchored to respondent's area]

Closing (text, included with the survey)

Thank you. Brent will synthesize responses across the team and bring patterns forward in the Sprint One diagnosis and the workshop preparation. If anything you wrote prompted a longer thought, you can email it directly to Brent or surface it in your calibration session with him later in Sprint Two.

Taxonomy balance 7 questions audited 0 anti-patterns flagged 6 taxonomy types covered

Coverage: Sequence (Q1, Q3), Specific Examples (Q2, Q4, Q5), Exceptions (Q4 branch), Relationships (Q6), Open-framework (Q5), Projection anchored (Q7). Opinion-only stance dissolved.

Framework-priming on Q5: The (a)/(b)/(c) scaffolding matches the three-way tagging rubric from the Operational Task Mapping module. Resolved with an explicit escape hatch ("if none of (a)/(b)/(c) feels right, describe the relationship in your own words") so respondents who don't fit the frame are not pushed into it.

Discovery Interview Guide

CEO and executive discovery interviews · a bank to draw from, not a script to run · the red flags and next moves are the facilitator's, never shared with the client

Use a few questions per interview and follow the threads they open. Asking all of them turns the conversation into an interrogation and produces shallow answers. Lead with the openers, which land at any maturity. Reach for the deeper probes only with a client already deploying AI, which you will usually know from the pre-contract conversations.

Openers (work at any maturity)

Posture, current footprint, and pain. They ask what the person sees and feels, not whether a deployment is failing. With an early-stage client, these may be the whole interview.

• Where do you see AI in the work today, sanctioned or not?
• What is the most repetitive, time-consuming, or error-prone part of your week?
• Where does AI worry you?
• What about it interests you?

Mission probe (only when a client leads with its values)

Skip this with most clients. For a credit union that leads with its mission, one open-ended question surfaces where AI meets what they stand for. Keep it generative. Do not hand them a values framework. Let them define it, and let your read of the room carry it from there.

• What does your credit union stand for that a member would feel, and where could AI strengthen or weaken that?

Deeper probes (when the credit union is already deploying)

These assume enough deployment to feel friction. With a pre-deployment client, skip them or turn them forward ("here is what to watch for as you scale"), because asking them backward presumes a failure that has not happened. Pick the block that matches what the leader is describing, then pick a couple of probes. Do not march the list. Each probe has a red flag (the comfortable non-answer that tells you to push) and a next move (a prompt to open the real thing without putting them on the defensive). The next moves are prompts, not a script. You will often have a better one in the moment.

"We can't point to anything that's actually better."

Q. What actually changed about how the work gets done since these tools came in? The steps, the handoffs, who decides what.

Red flag: "We trained everyone on the tools." That is adoption, not a change in how the work runs.

Next move: "Walk me through one task start to finish, the way it ran before and the way it runs now."

Q. Can you point to one thing that got better for a member or the bottom line because of AI, beyond being faster?

Red flag: "We're saving about ten hours a week." That is time, not a result.

Next move: "Where did those ten hours go? What is getting done now that wasn't before?"

Q. When the tool produces something (a draft, a score, a recommendation), what happens next? Does someone weigh it, or does the work just move forward?

Red flag: "It's our starting point and we refine from there." Often "refine" means fixing the wording without asking whether the answer is right.

Next move: "Tell me about the last time it was wrong. Who caught it, and how?"

Q. Who owns whether the tool's outputs are still accurate a year from now?

Red flag: "The vendor set it up," or "IT manages it." Running the tool and checking its outputs are two different jobs.

Next move: "If the model started drifting next quarter, who would notice first?"

"We're losing the judgment that made us good."

Q. If the tool went dark for a month, could your team still do the work at the level they held two years ago?

Red flag: any hesitation.

Next move: "Could the newest person on that team do it without the tool? What would slip?"

Q. What are your newer people learning that the AI can't teach them?

Red flag: "They're great with the tools." Tool skill has a short shelf life.

Next move: "When your best lender retires, what did the tool keep the next person from ever learning?"

Q. Which parts of the work that people used to find meaningful does the AI now handle?

Red flag: "It frees them up for higher-value work."

Next move: "Name the higher-value work. Is someone actually doing it this week?"

Q. Has anyone mapped which tasks the AI does well and which ones it doesn't?

Red flag: "People figure it out by trial and error." That is the path skill erosion takes.

Next move: "Where has trial and error already cost you? A wrong call that shipped."

"People are trusting it without checking."

Q. What does a wrong AI output look like in your shop? Could your team describe one?

Red flag: "It's usually pretty good." That is average performance. If no one can name the ways it fails, no one is catching the failures.

Next move: "Walk me through the last one that got through. What did it cost?"

Q. When the AI says something that cuts against what an experienced person believes, what happens?

Red flag: either "we go with the AI, it has more data," or "we ignore it, we know our members."

Next move: "When was the last real disagreement, and how did it actually get settled?"

Q. Who is responsible for knowing whether the recommendations are still right as conditions change?

Red flag: "IT handles the tools." Managing the tool is not validating what it produces.

Next move: "If you suspected an output was off tomorrow, who would you call, and would they know how to check?"

"We're nodding along too fast."

Q. When was the last time a meeting landed somewhere different from where the AI's answer pointed?

Red flag: "We're all looking at the same data." Same data should still produce different reads from different people.

Next move: "When did someone last talk the room out of the AI's answer? What happened?"

Q. Is it anyone's actual job in the room to argue the other side?

Red flag: "We encourage people to speak up." Encouragement is not a role.

Next move: "Who in the room is comfortable being the one who's wrong?"

"More data, no sharper decisions."

Q. For the last big decision, what did someone in the building know that never made it into the room?

Red flag: "We had what we needed."

Next move: "Who didn't you talk to that you would talk to if you ran it again?"

Q. When work moves from one team to the next, what gets lost in the handoff?

Red flag: "We have a good handoff process."

Next move: "Take the last handoff. What did the receiver have to go back and ask for?"

Q. Where do different functions actually argue a problem out together, instead of reviewing it one after another?

Red flag: "Each function signs off at its stage." Signing off is review, not synthesis.

Next move: "When did two functions last change each other's minds in the same room?"

Listening checklist (facilitator's, not for the client)

Keep this in front of you across every interview. For noticing and capturing in the room, not for routing what you hear to a finding. The dot-connecting happens after, across all the interviews, with your judgment. Mapping a symptom to a diagnosis live would prime you to hear what the map predicts and miss the connection that does not fit, which is usually the one that matters.

1. Anchoring on the AI. They take the model's recommendation as the answer. Sounds like: "the system flagged it, so that's what we did."
2. The AI agreeing with what they already believed. Treated as proof, not as a check. Sounds like: "the AI backed up our read."
3. Plausible but wrong, with no one checking. No one can describe a wrong output, and there is no verification step. Sounds like: "it looked right, it was well written."
4. Experienced people getting rusty. The work is harder without the tool than it used to be. Sounds like: "I used to do this in my sleep."
5. Newer people who can't work without it. They can't do the core task from scratch. Sounds like: "I've never built one without it."
6. No map of where AI fits. The tool gets used everywhere, by convenience rather than fit. Sounds like: "we use it for everything, it's easier."
7. Decisions quietly moving to the tool. Acceptance is near total and no one quite owns the call anymore. Sounds like: "we just go with what it recommends."

Provenance: the openers are the CU-native discovery questions, with the Clearwater-specific Six Principles question dropped. The probes, red flags, and next moves are adapted from Superadditive's questions experiment, reworked for credit unions. The research threads under a few red flags (experienced workers gaining least, automation dulling skill, handoffs losing context) are listening cues only, to verify against the primary sources before any of it reaches a client.

Foundation Assessment (4 Pillars)

Source: McKenna AI Strategy Canvas, Foundation Assessment

CEO Interview Guide 60 min, 1:1 with the CEO · structured across the 7-stage interview arc

A real-time conversation covering the CEO's AI posture, the CEO's current view of Septapod's AI footprint, the institutional identity that constrains what AI can do here, and the CEO's projection of where this work lands. The most useful interview signals come from responses that trigger an "a-ha" reaction or a "that's ridiculous" dismissal. Write those down verbatim. The guide is a flexible reference, not a script.

DISC awareness: Customize per CEO. Match the CEO's pacing. Move to substance fast, but do not skip the threshold and kick-off entirely.

Stage 1-2: Opener (scripted, 2-3 min)

"Thanks for the time. I want to use this conversation to understand how you think about AI at Septapod, how it fits with what makes Septapod Septapod, and where you want this work to land. Sixty minutes is the rough envelope. There are no right or wrong answers; the questions are designed to surface specifics rather than ratings. I will take notes as we go, and I will follow up on anything that needs more time after we synthesize this with the team survey and the function work. Sound good?"

Logistics: confirm recording consent. Note any hard stop. Adjust on the spot if time is shorter than expected.

Stage 3: Kick-off (3-5 min)

Q1 [kick-off]. How long have you been thinking about AI as a strategic priority for Septapod? When did it first show up on your radar?

Probes: "What changed at that point that made you start paying attention?" / "What were you reading or watching when it first felt important?"

Stage 4: Accept the Awkwardness (concrete, 8-12 min)

Q2 [sequence + complete-list]. Walk me through everything you know about AI showing up at Septapod today, sanctioned or otherwise. Tell me what's running.

Probes (use only if the CEO runs dry; do not list verbatim up front): "What else?" (ask twice) / "What about vendor products?" / "What about tools your team uses day to day?" / "Shadow use anywhere?" / "Where is the Copilot rollout right now?" / "Where do you suspect AI is showing up that you don't have visibility into?"

Q3 [specific example]. Tell me about a recent moment when AI work at Septapod either moved forward unexpectedly or stalled. What was happening?

Probes: "What did you do?" / "What did that tell you about where this work currently sits?"

Transition: "Okay. I want to shift to some questions about Septapod itself, separate from AI for a minute. Sound good?"

Stage 5: The Tipping Point. Core Questions (25-30 min)

Q4 [specific example, sacred cows]. Think of a moment in the past few years when someone proposed changing something at Septapod and you knew immediately it was off the table. What was it, and why was it untouchable?

Probes: "Who proposed it?" / "What did you say in the moment?" / "Is there a category of decisions where that comes up more than others?"

Q5 [code words, paradigms]. Tell me about a time when a new leader or new hire at Septapod did something that violated an unwritten rule without realizing it. What happened, and how would you explain that rule to someone joining tomorrow?

Probes: "What rules do you find yourself explaining most often?" / "Where do those rules come from?"

Q6 [compare across time, past events]. What past events at Septapod or in your career hold lessons for what's about to unfold with AI? What's the closest analogue you've watched play out?

Probes: "What did you take from it?" / "What would you do differently this time?"

Q7 [cross-functional sequence]. When something at Septapod requires multiple departments to coordinate on a decision or an initiative, what does that actually look like? Walk me through a recent example.

Probes: "Who tends to lead?" / "Where does coordination break down most often?" / "Where does AI fit into that picture today?"

Q8 [relationships, stakeholders]. Beyond yourself, who are the two or three people whose buy-in is most load-bearing for this AI work to succeed? What do you know about where they currently stand?

Probes: "What would change their stance?" / "Where do they disagree with you, if anywhere?"

Transition: "Okay. I want to step back to the bigger picture for the last stretch."

Stage 6: Reflection and Projection (10-12 min)

Q9 [projection, legacy]. When you move on from this CEO role, what do you want to leave behind that's tied to the AI work you're starting now? What would tell you, in five years, that this period was a success?

Probes: "Whose judgment of that would matter most to you?" / "What's the failure mode you want to avoid?"

Q10 [ideal state, 18-month horizon]. If Septapod's AI work goes the way you most want it to go, what is true 18 months from now that is not true today?

Probes: "What would a coworker notice?" / "What would a board member notice?" / "What would a member notice (if anything)?"

Q11 [exception, failure mode]. If Septapod's AI work stalls or goes sideways, where do you think it is most likely to break down, and what will be in the way?

Probes: "What would you watch for as an early signal?" / "Have you seen this break down at Septapod or elsewhere before?"

Stage 7: Soft Close (3-5 min, doorknob aware)

Closing questions:

• "Is there anything we did not cover that you think is important for me to know before I synthesize this work?"
• "Anything you want to ask me about how I am approaching this, or where I see the engagement landing?"
• "When I come back with the diagnosis after Sprint One, is there anything you want me to be sure to listen for in the team survey responses or in the function work?"

Doorknob awareness: Keep notes open and engaged until the conversation has fully ended. The most important data of the interview often surfaces in the last few minutes as the CEO feels the formal structure dissolving. If the CEO starts a new thought as you are closing the laptop, sit back down. Do not rush the departure.

Taxonomy balance 11 substantive questions 0 anti-patterns flagged 7 taxonomy types covered

Coverage: Sequence (Q2, Q7, kick-off probe), Specific Examples (Q3, Q4, Q5), Compare Across Time (Q6), Relationships (Q8), Code Words / Native Language (Q5 folded in), Projection Anchored (Q9, Q10), Exception (Q11), Complete List (Q2 folded with sequence).

Arc structure: All 7 stages of Portigal's interview arc present. Threshold, restating objectives, kick-off, accept-the-awkwardness, tipping point, reflection-and-projection, soft close. Transitions between sections are scripted.

Q2 scaffolding moved to probes: The original draft listed possible AI surfaces in the question stem. Live-interview risk: the interviewer's voice fills the silence the participant needs to think. Resolved by moving those categories from the question stem to the probe list; they get used only if the CEO runs dry.

Governance Readiness Assessment Parallel fast track, ships end of Week 2

The AI Policy (board-approved earlier this year) already defines decision authorities: Board approval triggers, Executive Team implementation authority, VP of Risk and Reporting for risk assessments, Tech Steering Committee for high-risk approvals, AI Taskforce for culture and training, AI Solutions Use Questionnaire as intake. This assessment tests whether those authorities are operational, not creates new governance from scratch.

Has the Tech Steering Committee reviewed an AI vendor through the defined process? Has the AI Solutions Use Questionnaire been used for a real request? Has the AI Taskforce met? Who is on it? Where do the AI Policy's authority structures work as written? Where do the AI Policy's authority structures need operational support?

Why this ships fast: Without confirming that existing governance authorities are operational, Sprint Three pilots stall on "who can sign off on this?" The assessment identifies what works and what needs support before pilots begin.