# Interview Guide Question bank for the two interview phases. Small batches (3 to 5 questions per turn), reflect answers back, never a form dump. ## Phase 1: Company intake (documents first, then gaps) ### Step 1: Ask for source material before asking questions Open by requesting whatever the executive can share: 1. Documents: annual report or 10-K, investor deck, strategy memo, org chart, prior AI/digital assessments. PDFs can be read directly. 2. Website: the company URL. Fetch it (homepage, about, products/services, investor relations, careers pages) with the available web-fetch tools and mine it for facts. If no web tool is available in the session, say so plainly and ask the executive to paste the relevant pages. 3. Public-company shortcuts: for listed companies, the 10-K's Item 1 (Business), Item 1A (Risk Factors), and MD&A sections are the densest sources. Read them. 4. Web research beyond the sources: search for reported financials, employee counts, competitors, announced AI initiatives, and executive commentary that the documents and site do not cover. Use any connected data sources (Claude connectors) available in the session as well. Extract before you ask. Every question you can answer from the documents or from research is a question the executive should not have to answer. Tell them what you learned, with the source named, and have them confirm or correct it; correction is faster than dictation and demonstrates you did the reading. Where research and documents both come up empty, tell the executive exactly what could not be found online so they know it is theirs to supply. ### Step 2: Profile fields to fill (from documents where possible, interview for the rest) - Company name, HQ, founding year, ownership (public/private/PE-backed/family) - Industry/sector and sub-segments; what the company actually sells - Business model: who pays, for what, how often - Scale: revenue, employees, geographies, customers; revenue per employee (derive) - Segments/divisions and rough revenue mix - Key competitors and market position. Competitors carry a hard gate: research the competitive landscape first and propose the likely competitive set, with a line on each (positioning, scale, recent moves). Present it for the executive to verify or revise, and do not move forward until they confirm the list. If research finds no clear competitors, say so and ask them to name the set. - Regulatory regimes that govern the business - Technology estate: core systems (ERP, CRM), cloud posture, data platforms if known - Current AI posture: anything already deployed, announced, or piloted (check the website and filings for AI mentions; companies often disclose more than the exec team remembers) - Strategic priorities as stated by leadership (from filings/site) vs. as felt by the executive ### Step 3: Confirm the profile Render the company profile HTML deliverable, present it, and loop on revisions until approved. Do not proceed to the qualitative interview with an unapproved profile. ## Phase 2: Qualitative interview (where the transformation stands) Open-ended. Listen for specifics; push past generalities ("we're experimenting" is not an answer; ask what, who, and since when). Cover: ### Ambition and mandate 1. What does a successful AI transformation look like for this company in three years? What breaks if it doesn't happen? 2. Who owns the transformation today? Is there a named leader with budget, or a committee with opinions? 3. What has the board actually asked for: a strategy, a pilot, cost savings, or nothing? ### Current state 4. What AI is genuinely in use today, by whom, and how often? What was tried and abandoned, and why? 5. Where is the shadow AI: what are employees using without permission or governance? 6. What was the last AI decision that stalled, and where did it stall (legal, IT, budget, fear)? ### Organization and culture 7. Which function is most eager, and which is most resistant? What is the resistance actually about? 8. What happened to the last major technology change program (ERP, cloud, digital)? The AI transformation will inherit that muscle memory. 9. Where does the talent stand: who can build, who can buy well, who can govern? ### Constraints and assets 10. What data does the company have that competitors don't? Is it accessible or trapped? 11. What are the non-negotiables: regulatory lines, union agreements, customer promises, brand positions? 12. What budget realistically exists for the next 12 months, and who signs? Synthesize this phase back to the executive in a short "where you stand" paragraph and get their agreement before moving to the pillars. That paragraph goes into the Strategic Outlook verbatim. ## Phase 3: Pillar questioning patterns Use mixed question formats deliberately; vary them so the conversation doesn't flatten into a survey. For every pillar, teach the pillar first (one short paragraph plus the buckets), then question. ### Pillar 1: Application Domains (the Where) - RANKING: rank the four domains by opportunity size for this company over the next 24 months. Then rank them by where the company is spending attention today. The delta between the two rankings is strategy. - SINGLE CHOICE: which single domain is the beachhead for the next two quarters? - OPEN: for the top two domains, what are the two or three concrete use cases? (Seed with domain use cases from the framework and anything surfaced in Phases 1 and 2.) - MULTIPLE CHOICE: which domains are already seeing unofficial/shadow AI usage? ### Pillar 2: AI Capabilities (the What) - For each priority domain, MULTIPLE CHOICE: which capabilities apply (Automation, Augmentation, Analytics & Predictive, Generative)? Enforce the RPA boundary on Automation answers. - RANKING: rank the four capabilities by expected value for this company overall. - OPEN: for each chosen domain-capability pair, what would the flagship use case be? Capture as a one-line triple-in-progress. - SINGLE CHOICE: which capability is the organization most overconfident about? (This question surfaces hype exposure; use it.) ### Pillar 3: Implementation Strategy (the How) - For each flagship use case, SINGLE CHOICE: where on the spectrum does it belong (Ready-to-Use SaaS, Embedded AI, Configured/Fine-Tuned, Custom/Ground-Up)? Challenge answers that default to the extremes: everything off-the-shelf means no moat; everything custom means no delivery. - MULTIPLE CHOICE: which embedded-AI features does the company already pay for but not use (Copilot, Einstein, SAP AI, etc.)? These are free wins. - SCALE 0-10: appetite for vendor dependence, and appetite for building internal AI engineering capability. - OPEN: what proprietary data or process would justify a Configured or Custom build? Record every answer with its question type; the Strategic Outlook renders rankings as ordered lists, choices as selections among the four buckets, and open answers as narrative.