# Rewrite memo — data-viz **Word count (body):** ~1,080 words · **Hero variant:** V2B · **Anonymisation:** Tier A (with explicit synthetic-data disclosure) --- ## 1. The story I found The case's argument is clear and unusual for a portfolio: this is a case about a method, demonstrated live, with yourself as the subject. The through-line a reviewer should remember: the shift from chatbot to agent doesn't change the output quality (much) — it changes *where the work happens* and therefore what skills matter. The bottleneck shifts from execution to articulation, which happens to be exactly the skill service designers spend years building. That argument is present in the existing case in the "What's actually different" section, but it's surrounded by so much round-by-round process description that it gets diluted. The rewrite moves the argument forward, treats the two rounds as chapters that build toward the insight rather than a blow-by-blow transcript, and names the reframe explicitly. The secondary argument — this is a note to Harmonic colleagues, not a technical note to analysts — gives the case its specific audience and specific stakes. That framing is present in the InsightCallout in the original. The rewrite promotes it to the argument paragraph. --- ## 2. What I kept - **"I have still never opened a Python file."** Too specific and funny to cut. Kept in Chapter 03. - **The three round-one moments** — bubble chart broken, bubble hidden, Leads-absorb-everything story that dissolved on contact. All three are kept in Chapter 02; they're the case's strongest evidence that round one has real, specific failure modes. - **The brief to Claude Code pull-quote** — "Audience is leadership. Use whatever charts and structure..." — kept in Chapter 03; it's the clearest illustration of how round two felt different. - **"I did not write the HTML. I wrote the brief."** Kept verbatim as Chapter 04's closing beat. This is the case's single sharpest sentence. - **The three pushes** (grouping, borderline cases, tone) — kept in abbreviated form in Chapter 03; they show iteration at conversation speed, which is the round-two experience. - **The synthetic dataset disclosure** — kept; it's ethically important and should be in the metadata strip or the argument paragraph's opening. - **The interactive site and its public URL.** Kept in the asset-needs as the primary visual for Chapter 03. --- ## 3. What I cut and why - **The full GPT conversation transcripts** for both rounds. The exchanges were illustrative but long. The key moments (chatbot asks for context → gives a plan; agent reads directly → acts) are captured in two-sentence descriptions. The full back-and-forth is replaced by a Placeholder for the relevant visual (round 1 crummy chart; round 2 interactive output). - **"What I'm taking from this"** (3 lessons-learned bullets: brief questions not chart types; verify when surprised; stop when it tells the story). Lessons-learned framing. The substance is absorbed into Chapter 04 and the reframe section. - **"What I'm still thinking about."** Partially cut. The synthesis-to-action gap observation is genuinely forward-looking and carries content; it's in the scaling section as "whether leadership acts on the open questions." The "how much of this is portable to client work" meditation is compressed into two sentences in the scaling section. - **Detailed InsightCallouts mid-case.** The original had three inline InsightCallouts. In the rewrite these become either argument paragraph content, reframe section content, or are removed — the V4 brand system has InsightCallout components for the live site, but in the MDX source these work better as inline beats within the chapter text rather than floating callouts that interrupt the read. - **The `` interactive component embed.** This is in the live case because it's the live Astro site. In the MDX source for the rewrite, it's replaced by a Placeholder pointing to the deployed URL for Whitney's review session, with a note that the interactive version can be embedded in the live MDX. --- ## 4. What I added - **Argument paragraph.** The existing case opened with "The synthesis was the work. The experiment was what I wanted my colleagues to see." That's close to good but it's two sentences, not a thesis. The rewrite argument paragraph names the case's argument (the working relationship changed, not just the outputs) and its audience (colleagues, not the general public) before the case starts. - **Chapter 04 as the synthesis chapter.** The existing case had "What's actually different" as a section that came after the two rounds but before a pull-quote section and the "three pushes." The rewrite promotes the synthesis to its own chapter and builds the bottleneck-shift argument fully rather than as a series of observations. - **Formally named reframe section.** The "I expected better outputs, but the relationship changed" insight was present in the InsightCallout in the original. The rewrite makes it the named reframe. - **Formally named scaling section.** "What I'm taking from this" + "What I'm still thinking about" → a scaling section that makes a specific claim about portability to client work and names the within-Harmonic adoption gap explicitly. --- ## 5. Hero variant choice: V2B (typography-led) "What changes when the AI shares your environment" is a title that carries the argument directly. The case's insight is a single claim (the bottleneck shifts from execution to articulation) that's best served by a strong header-and-sub-banner pairing rather than a spine that previews a multi-chapter sequence. The two rounds are chapters, but they don't form an arc that a reviewer needs to see in the hero to follow the case — they're two parallel illustrations of the same point. V2B lets the title do the work. --- ## 6. Open questions for Whitney 1. **The deployed deliverable URL.** The live case references `https://pain-points.whitney-masulis.workers.dev` as the deployed interactive site. Is this still live? If the URL has changed or the site has been taken down, the asset-needs item and the scaling section reference need updating. 2. **The Harmonic colleague who ran the upstream synthesis.** The existing case says "a colleague and I" without naming the person. For attribution and credit, should the colleague be named (per the content strategy's guidance on crediting collaborators)? Even "a senior Harmonic researcher" would be more specific than "a colleague." 3. **The studio Spotlight recording.** The existing raw-case-material contains a Spotlight transcript (`Spotlight custom GPT ERIE Aging Carestar_otter.ai.txt`) — but that's for the agetech case. Is there a parallel Spotlight transcript for the data-viz case? A quote from a colleague's reaction during the Spotlight would be the case's missing "moment the reframe landed" beat.