The AI-augmented analyst
What AI actually changes about the BA role (and what it can't touch), the working setup you'll use all course, and the investigation you've just been handed.
A business analyst's job has three layers: gathering (interviews, documents, data pulls), sense-making (what does it mean, what should we do), and persuading (getting the organization to act). AI has become genuinely excellent at the first layer, usefully good at drafting the third, and — this is the part vendors won't tell you — still entirely dependent on you for the second. This course is about using AI hard on layers one and three so you have more time and better raw material for the layer that is actually your job.
- AI is a tireless junior analyst: it drafts interview guides in seconds, synthesizes fifty pages of notes without fatigue, writes the first version of every document, and never complains about reformatting. Like a junior, everything it produces needs review before it leaves your desk.
- AI is not accountable. When the recommendation is wrong, the room looks at you, not the model. That asymmetry drives every discipline in this course: verify numbers, ground claims in quotes, separate what the data shows from what you infer.
- The analysts who lose to AI are the ones who only did layer one. The analysts who win with it are the ones who use the freed-up hours for judgment — the layer AI can't do and stakeholders can't skip.
Your assignment (it arrives Monday morning)
You are Maya Torres, business analyst at Harbor Lane — a specialty coffee retailer with a web store and three physical locations (Harborfront, Midtown, Station). The COO's message is two sentences: "Refund costs are up about 40% versus last quarter and nobody can tell me why. Can you own this — cause, cost, and a recommendation — by end of month?" That's the whole brief. Vague sponsor, real money, hard deadline: the natural habitat of the business analyst. Every lab in Modules 1–4 advances this investigation, and by Module 5 you'll run the same playbook on a question from your own job.
Your working setup
- An AI assistant with long context (Claude or equivalent) — you'll paste transcripts and data tables into it constantly. If your organization has an approved tool, use that one; the prompts here are tool-agnostic.
- One project/folder per investigation. Keep the brief, notes, transcripts, and drafts in a dedicated AI project (or one long-running conversation) so the assistant carries context — re-explaining the investigation in every chat is the #1 rookie time-waster.
- A spreadsheet you're comfortable in. AI will help you analyze, but you verify in the spreadsheet. No code in this course; if you can pivot-table, you're equipped.
Know your company's AI data policy before pasting anything. In this course the data is fictional. At work: internal business data is usually fine in an approved enterprise AI tool and never fine in a personal account; customer PII should be masked either way (initials, IDs). The AI Foundations course's four-class data framework applies to every paste you'll do here — when in doubt, ask before pasting.
This is the Business Analyst track's core course. It assumes AI Foundations (prompting basics, verification habits). Where your investigations need trustworthy data infrastructure, that's Data Foundations for AI territory; where they need a governed way to query metrics conversationally, that's the Conversational Analytics Agent course — both are cross-referenced where relevant, neither is required.