Case Study 1: Elena's Domain Sprint — From Zero to Expert Brief in One Week

The Engagement

On a Monday morning, Elena accepts an engagement she almost declines. A healthcare technology company — a former client from a different project — asks her to evaluate their strategic options for entering the behavioral health market. The CEO wants a preliminary strategic brief by the following Monday, ahead of a board strategy session.

Elena's problem: she knows almost nothing about behavioral health. Her healthcare work has been in medical devices and hospital operations. Behavioral health is a different regulatory environment, a different payer structure, a different clinical model, and a different competitive landscape. She has seven days, two other active engagements, and an evening speaking commitment midweek.

The old approach would be to decline or ask for more time. A week is not enough time to develop genuine domain expertise in a new healthcare subsector — or it was not, before AI.

Elena accepts the engagement. She has a protocol for domain sprints now. She has tested it four times in the past year, always in adjacent domains. This is the most aggressive test she has attempted.

Day 1: Mapping the Territory

Elena begins at 7 AM with a two-hour AI orientation session. She works through a structured sequence of "teach me" conversations across six topic areas:

Session 1: System architecture. She asks Claude to teach her how the US behavioral health system is structured — who provides care, who pays, how reimbursement flows, and what the regulatory bodies are. She asks follow-up questions for every concept she does not understand immediately. She learns about the fragmentation between mental health and substance use treatment, the role of managed behavioral health organizations, and the differences between commercial insurance and Medicaid behavioral health coverage.

Session 2: Regulatory environment. HIPAA, 42 CFR Part 2 (substance use disorder record confidentiality), mental health parity requirements, and the regulatory implications of digital health. She asks specifically about what has changed in the last three to four years.

Session 3: Digital therapeutics landscape. What products exist, what evidence supports them, what FDA pathways apply (Software as a Medical Device vs. wellness tools), and what reimbursement pathways exist.

Session 4: Payer dynamics. She asks about how behavioral health reimbursement differs from medical reimbursement, the managed care organization landscape, and what employers are doing with behavioral health benefits.

Session 5: Competitive landscape. Who the major players are in digital behavioral health — the established companies, the well-funded startups, and the models that have failed.

Session 6: Market sizing. How large the market is, what the growth drivers are, and what market sizing methodologies different analysts use.

After each session, Elena does two things: she writes a brief summary of what she has learned in her own words, and she identifies claims she needs to verify and sources she needs to read. By the end of the two-hour orientation, she has a twelve-item reading list.

She spends the rest of Day 1 tracking down the reading list items. Six are accessible immediately (government websites, free reports, open-access papers). Three require her firm's database subscriptions. Three she cannot find — two may not exist, one may be behind a paywall she does not have access to. She makes substitutions.

Days 2 and 3: Primary Source Reading

Elena's reading list has been shaped by her AI orientation. Instead of starting with broad behavioral health overview reports and working toward specificity, she starts specific. Her orientation has already given her the overview; what she needs is the depth and verification.

She reads: - The CMS mental health parity final rule summary and the enforcement guidance document - A 2023 Health Affairs analysis of digital mental health reimbursement coverage - The FDA's Software as a Medical Device (SaMD) action plan - A meta-analysis of smartphone-based mental health interventions from JAMA Psychiatry - A market intelligence report on the behavioral health technology space - A competitor's SEC filing (the company is publicly traded and provides detailed market context)

She takes structured notes after each source: key findings, specific data points with page references, claims that confirm or contradict her AI orientation, and questions the source raises.

Three times during reading, she discovers that her AI orientation was incomplete or misleading. The CMS mental health parity rules are more complex than the AI described, with significant recent litigation activity that changes the practical landscape. The digital therapeutics reimbursement picture is more heterogeneous than AI suggested — it depends heavily on whether the product is prescription-only or direct-to-consumer. She updates her notes and flags these corrections.

Day 4: Expert Interviews

Elena had prepared her expert interview questions using AI: "Given what I know about the behavioral health market [she summarizes her current knowledge], what questions would be most valuable to ask two experts: one with a healthcare policy background and one with a digital health startup background?"

The AI-suggested questions are organized into four categories: regulatory landscape questions, market dynamics questions, competitive positioning questions, and go-to-market questions. Some are obvious; a handful are genuinely insightful and reflect connections across the domain that Elena had not explicitly formulated yet.

She interviews a healthcare policy specialist (a professor at a local university who has done behavioral health consulting) and a former COO of a behavioral health technology company. Both interviews are one hour.

The policy specialist's most important contribution: the enforcement landscape for mental health parity is shifting significantly, and companies entering this market in the next two to three years are entering a period of increasing regulatory scrutiny. This nuance did not appear clearly in any written source Elena had read. It changes her risk assessment materially.

The startup COO's most important contribution: two specific companies that Elena had flagged as major competitors in her AI orientation have fundamentally different business models than AI described — one has pivoted from B2C to employer benefits, and the other has essentially exited the direct-to-consumer market. Her competitive landscape understanding is now significantly more accurate.

Day 5: Synthesis and Brief Writing

On Day 5, Elena writes the strategic brief.

She submits her reading notes (not original sources, not AI summaries — her own notes) to Claude for synthesis assistance. The synthesis prompt:

"Based on my research notes below, identify the major themes, any contradictions between sources, and the key uncertainties that remain. Work only from the material I have provided. For each major theme, indicate which sources support it."

The synthesis is useful but imperfect. Claude's synthesis slightly overstates the consensus on digital therapeutics efficacy — the literature is actually more qualified than the synthesis suggests. Elena catches this because she has read the papers. She adds the appropriate qualification herself.

She writes the brief in five sections: 1. Market overview and sizing (incorporating verified statistics from her reading) 2. Regulatory environment and risk assessment (significantly updated by her expert interview) 3. Competitive landscape (significantly updated by her expert interview) 4. Digital therapeutics evidence base (synthesized from her primary source reading) 5. Strategic options (three options with risk/opportunity assessment)

She writes Section 5 — the strategic options and recommendations — entirely herself. This is the section where her judgment and her client relationship matter most. AI drafts an outline that she discards. The strategic options reflect her assessment of the client's capabilities, risk tolerance, and market timing — none of which AI can know.

The Result

Elena delivers the brief on Sunday evening, a day ahead of the deadline.

The CEO reads it Monday morning before the board session. His reaction: "This is exactly what we needed. How long have you been following this space?"

She has been following it for seven days.

More importantly: the brief is accurate. In the week following the board session, the CEO asks his internal team to fact-check the specific market figures Elena cited. All are traceable to identified sources. The regulatory nuance from her expert interview proves to be the most valuable element of the brief — it shapes the board's risk framing in ways that affect the decision substantially.

What the Domain Sprint Protocol Requires

Looking back on the engagement, Elena identifies the non-negotiable requirements that made the sprint work:

Reading cannot be skipped. The AI orientation is valuable for directing the reading, but three of Elena's most important insights — the parity rule complexity, the digital therapeutics reimbursement heterogeneity, and the startup COO's competitive corrections — came from sources she read herself, not from AI orientation. If she had relied on AI orientation without reading, her brief would have had significant errors.

Expert interviews fill the gap that written sources cannot. The regulatory enforcement shift and the competitive pivots were not in any written source Elena found. They were in the heads of practitioners. AI and published sources together do not cover what experts who are living in the domain know.

Verification is continuous, not terminal. Elena does not run a verification pass at the end. She verifies as she reads — checking AI orientation claims against primary sources as she encounters them. By the time she writes, she already knows which claims are verified and which are uncertain.

AI writes the orientation but not the judgment. The strategic options — the section where Elena's client is actually paying — are written without significant AI input. AI generated an outline she discarded. The professional judgment is irreducibly hers.

The Time Accounting

Elena estimates the engagement required approximately thirty-five hours of work over five days. She estimates that without AI assistance, the same domain sprint — if achievable at all in a week — would have required fifty-five to sixty hours and might still have been less thorough, because the AI orientation allowed her reading to be highly targeted.

The time saved is real. But where it was saved is instructive: AI compressed the orientation and structural synthesis phases. It did not compress the reading, the interviews, or the strategic judgment. Those take the time they take.