Current state, verified July 24, 2026. The live teaser returns a full public HTML page titled "Thermo Fisher / mySupply — 0→1 Pharma Supply Chain, $20M+ Margin — Juno Chen." The hero leads with the delivery-gate cost problem. Headline metrics appear in the hero section. The body includes an interactive app showcase, a mobile batch-detail mockup, a fourteen-card design-system bento grid, and a closing section with three pillars. Two CTAs labeled "Request the full case study" link to cs-02-supply-chain.html. The irreversibility frame is present. The agentic vision is absent.
The metadata is the strongest part of this page. The body has a structural problem: it eliminates the reason anyone would request the full case.
Check 1: Page title and metadata
Finding: Strong. Leave it.
Title leads with outcome proof: "$20M+ Margin." Meta description packs all four proof points into one line. The Open Graph description takes a complementary angle: "No platform existed. Six pharma partners, nine manufacturing sites, a $20M visibility gap. Designed from scratch. All six adopted it." A VP of Design seeing this in a Slack preview gets scale and outcome before clicking. A CPO gets business-case language before design language. Both buyers served in one tag.
No changes.
Check 2: First screen framing
Finding: Diagnosis-forward. One soft spot.
"The supply chain six partners never had" implies the gap existed before you named it. The hero intro does harder work: names the structural problem (partners learned about exceptions at the delivery gate, when the cost was already 3–5× higher), then names the result (one platform, all six adopted, zero to live in twelve months). A VP of Design scanning this understands you diagnosed the gap, not just filled a brief.
The role label is the soft spot. "Product Design Director" is accurate and generic. The closing section tells a story about a design leader who ran thirty-two interviews, identified five failure modes, and built a platform that displaced entrenched workflows across nine sites. None of that is visible in the hero. The role label should carry more of that weight for the buyer who never scrolls past the first screen.
Recommendation: Change the role row to "Product Design Director · Solo design lead, 0→1." Pre-answers the question every hiring manager asks at Director level: did she do the work or manage people who did. Some buyers will read "solo" as scope limitation. The hero intro corrects that within three seconds of scrolling, and the team row already shows the full build team (1 designer, 2 engineers). Specificity wins here.
Check 3: Outcome proof without scrolling
Finding: Three of four headline metrics visible. Sufficient.
The outcome row shows "$20M margin opportunity · 100% partner adoption · 83% IRR." The 42% overhead reduction lives only in the full case. Three metrics in the first viewport is enough. The 83% IRR is the one that stops a CPO mid-scroll. IRR is their language, not yours. Using it signals you understand what they measure.
No changes.
Check 4: Craft signal at scan speed
Finding: Strong. The artifacts are doing real work.
The app showcase with the desktop iframe and the mobile batch-detail mockup lands immediately after the hero. The mobile mockup shows a real batch: AB1234, Manufacturing, At Risk, 65% progress, sterility and particulates tests pending. That is a working system with domain-specific states, not a wireframe.
The bento grid goes further. Status badges with exception states (LATE, AT RISK, ON TRACK, HOLD, RESOLVED). KPI cards showing "AI Gate Reviews 24 this wk / 87% avg confidence." Regulatory-constraints cards explaining GDP and cGMP design consequences. This signals you designed within a regulatory framework, not around it.
No changes to the craft artifacts themselves. The density issue is structural, not qualitative. Check 5.
Check 5: Density
Finding: Too dense. The teaser is doing the full case's job.
Count what you're showing: an interactive app showcase, a mobile mockup with domain-specific detail, fourteen component cards in the bento grid, each with explanatory copy. The partner scorecard card explains that this was the first time all six partners saw each other's delivery performance. The exception-summary card explains which states cost money. The manufacturing-utilization card names a specific conversation that recovered $2M. The regulatory-constraints card walks through GDP and cGMP implications for every design decision.
A teaser proves three things and creates appetite for the rest: the problem was real, the design work was substantive, the outcomes were measurable. Everything beyond that belongs behind the gate. Right now a buyer who reads the teaser has already absorbed the design thinking, the system architecture, the regulatory reasoning, and the specific mechanisms. What would they request access for? They already know.
Recommendation: Cut the bento grid from fourteen cards to four. Keep:
- Status badges — craft signal
- KPI card with AI Gate Reviews metric — currency signal
- Regulatory-constraints card — domain depth
- One operational card (the exception summary or the partner scorecard, not both)
Remove the rest. They belong in the full case. The teaser should leave the buyer thinking "I need to see how she got from thirty-two interviews to five failure modes." Right now it leaves them thinking they already did.
Check 6: CTA
Finding: The label promises something the link cannot deliver.
Both CTAs say "Request the full case study" and link to cs-02-supply-chain.html. That page returns HTTP 200 with full HTML but runs a client-side script that redirects visitors unless localStorage.getItem("pf_access_v1")==="1". I flagged this in issue #3 as a P1 risk. The gating protects against crawlers but redirects humans with no explanation.
"Request" implies a process. There is no process. No form, no email gate, no contact mechanism. A buyer who clicks either gets silently redirected or, if they're technically inclined, finds the full HTML in the page source.
Recommendation, pick one:
Option A (preferred): Replace the CTA link with a mailto or Calendly link. Change the label to "Request access to the full case study." The click opens an email draft or a scheduling page. Real gate. Real conversion event.
Option B: Keep the direct link but fix the client-side gate to show a request form instead of redirecting. "This case study is available by request. [Your name] [Your email] [Send request]." The buyer sees the work exists and that access is intentional, not broken.
Do not leave the current state. A CTA that leads to a silent redirect is worse than no CTA at all.
The irreversibility frame
Present. Underleveraged.
The teaser uses the 3–5× cost language twice: once in the hero intro, once in the closing. The regulatory-constraints card adds the compliance dimension. What it never says is the word that makes this case land with buyers at Vanta, Headway, Ambience, and Rubrik: irreversible. The Trust essay says it directly:
"The downstream cost of a wrong automated release was irreversible and had a name on it."
The teaser says the cost was higher. The essay says the cost was permanent and personal. Those are different registers. The essay's register is the one that connects with buyers whose own products carry consequences that cannot be undone. A wrong credential issued at Vanta. A claim released at Headway. A batch shipped at Thermo Fisher. The pattern is the same across all three: once it's out the door, the cost is not remediation. It is regulatory, reputational, and named.
Recommendation: In the closing paragraph, after "where resolution cost 3–5× more," add one sentence: "A batch released wrong doesn't get un-released. The cost has a regulatory trail and a name on it." Twelve words. Connects the teaser to the Trust essay's language without quoting it. Tells a buyer at a compliance or healthcare company that you understand their stakes, not just their interface patterns.
The agentic vision — should it appear?
Yes. One line, not a section.
The full case closes with a detailed "If I were building this today" architecture: five continuous agents, one irreplaceable human QA-release gate, mockups showing agent activity and confidence scores. That level of detail would overload the teaser, especially after cutting the bento grid.
But the complete absence of the agentic frame is a missed opportunity. The Trust essay's §03 names this case explicitly as the example of "five agents running continuously" with "one human gate that was non-negotiable." In issue #4 I argued that the Trust essay makes the portfolio legible across current Director+ anxieties. The connection between CS-02 and that essay is the strongest thread in the portfolio. The teaser should pull it. Lightly.
Recommendation: Add one line to the closing section, after the three pillars and before the final CTA: "The full case includes a forward-looking architecture: five continuous agents replacing the pull-based model, one human gate that stays. The regulatory release signature is irreplaceable." Two sentences. Signals the case has a forward-looking dimension without requiring the buyer to absorb the full mockup. Creates a reason to request the full case that the current teaser does not provide.
Fix the CTA first. A broken conversion mechanism makes every other improvement academic. Then cut the bento grid to four cards, add the irreversibility sentence, add the agentic-vision line, and update the role row.
- Fix the CTA. Broken conversion makes every other improvement academic.
- Cut the bento grid to four cards. The teaser is currently eliminating its own reason to exist.
- Add the irreversibility sentence. Twelve words. High impact for regulated-industry buyers.
- Add the agentic-vision line. Creates a reason to request the full case and connects to the Trust essay.
- Update the role row. Small change. Preempts a common evaluative question.
Tighten the teaser so it creates appetite instead of satisfying it. Fix the CTA so the appetite has somewhere to go. Add the two sentences that connect this case to the irreversibility frame that makes it the strongest regulated-industry proof in the portfolio.
- Rubrik's reversibility language: Rubrik's AI launch describes autonomous actions as auditable, attributable, and reversible with human review when an action cannot be undone, which mirrors the exact irreversibility frame CS-02 should be sharpening for.
- Vanta's Head of Design scope: Vanta's live posting asks for agentic UX patterns, AI-native design practice, and launch review frameworks across a ~40-person org spanning GRC, Trust, Platform, Self-serve, and AI — the kind of buyer who would read CS-02's regulatory-constraints card as proof of domain fluency, not decoration.
- Headway's clinical workflow stakes: Headway's Design Director posting names a single provider session that touches scheduling, telehealth, clinical documentation, outcome measurement, insurance verification, and billing, making it the closest analog to CS-02's delivery-gate problem in a healthcare context.
- Client-side gating pattern persists: The full CS-02 case page still returns HTTP 200 with complete HTML before the
pf_access_v1redirect fires, a pattern first flagged in issue #3 that remains unresolved and directly undermines the CTA fix recommended above.