The Customer Brief
Define the scope of Lakeside Hospital Network’s AI-assisted clinical note drafting tool to reduce clinician documentation burdens while prioritizing accuracy and patient safety. Establish clear decision-making responsibilities among clinical leadership, privacy, legal, and operational teams. Ensure compliance with HIPAA requirements before processing patient data. Require clinicians to review and approve all AI-generated drafts before incorporating them into medical records. Avoid automatic filing to prevent inaccuracies from affecting patient care. Refine the tool’s functionality, establish appropriate safeguards, and verify regulatory compliance before proceeding with implementation.
We'll cover the following...
A clinical note is part of the medical record, and other clinicians read it to understand a visit and make later care decisions. That makes note drafting one of the most tempting healthcare uses of a language model and one of the easiest to get wrong, because a fluent draft can state a finding that nobody ever said. The final case in this course puts you in front of a hospital that wants exactly this assistant, and the interview questions that follow test judgment on data, safety, and when to say no.
Who is Lakeside, and what does it want?
Lakeside Hospital Network is a US hospital network. Its clinicians document each encounter in the hospital record after the visit, typing or dictating the note themselves. A clinical sponsor wants to cut that documentation burden by turning encounter material into draft notes.
The sponsor expects time savings, but Lakeside has no measurement of how long documentation takes today or how much a drafting assistant would save. Treat the saving as the goal to measure, not a result.
Three groups are affected by what you build. Clinicians write and sign the notes, and they answer for what the record says. Patients’ words are captured, their information moves through new systems, and later care depends on what the note records. And the record’s other readers, such as nurses, specialists, and on-call staff, act on a note without knowing how it was drafted.
Compared with the research assistant earlier in the course, the stakes are different in kind. An unsupported sentence in a research memo wastes an advisor’s time. An unsupported sentence in a filed note can change a later treatment decision.
Who decides what?
The FDE role covers discovery, technical scoping, system design, building, and rollout alongside the customer’s engineers and domain teams. At a hospital, your first discovery job is to find out who owns each decision, because most of them aren’t yours:
Clinical leadership: Defines the intended use, who reviews a draft, and who may file a note.
Privacy and legal: Decide what the law and hospital policy allow for recording, processing, and sharing patient data.
Security and engineering: Assess access, integration with the record system, and how the service runs.
Operations: Supports the workflow day to day, handles failures, and ...