Voting as0 of 15 calls made · saves live · teammates' dots appear after you answer

Scribe-X · New Homepage · Decision Page · July 2026

The Homepage Bake-Off

Two strong builds of the new Scribe-X homepage exist, and the final page should take the best of both. This page compares them section by section through six lenses, with the Magnetic Messaging Framework (V4, July 6) as the judge. Each contested section carries its decision question right where you're looking at the evidence. Pick your name in the bar at the top, answer as you scroll, and your calls save to the shared board instantly ... teammates' votes appear as colored dots once you've made your own call on a question. The answers shape the final build.

Build 1 · Oliver's

Dark, cinematic Next.js build. One hero, scroll-driven storytelling, a dedicated security section, and the Practice Health Score quiz embedded in the page.

Build 2 · Greg's

Light, narrative-first build. Carries the full MMF spine: the toggle-bar diagram in the hero, the buying committee, and the cost of standing still. (Hero headline vote pending ... it's the first decision in the Hero round.)

The Short Version

Both builds win ... in different categories. The final should merge them.

Scored across six lenses, the two builds finish nearly even overall, and each one is clearly stronger where the other is weaker. Greg's build carries the framework: the positioning spine, the buying committee, and the honest availability labels. Oliver's build carries the mechanics: cleaner page semantics, the security section, the embedded quiz, and the best single trust device on either page (the five vendor questions). Neither should ship alone when the merge is this obvious.

LLM readability

Greg

7.5

Oliver

7.0

How an AI answer engine reads the page. Greg's HTML serves real stat values to crawlers; Oliver's count-up animation serves initial zeros ("0.0–3.8 more patients"). Oliver has the single H1 and meta description. Neither page has JSON-LD yet ... the final needs it.

ADA / WCAG

Greg

5.5

Oliver

6.0

Both builds fail AA contrast today in fixable ways (11 fails vs 13). Oliver's semantics are cleaner; his scroll-driven problem section needs a reduced-motion fallback. Details in the appendix, including the greyscale-logo question.

The FQHC buyer

Greg

8.5

Oliver

7.0

Greg's build names and mirrors the buying committee (CMO, CMIO, CIO) so each buyer finds their own card. Oliver's speaks to a general "healthcare organization." The committee is the deal.

Conversion / CRO

Greg

7.0

Oliver

8.0

Oliver embeds the Health Score quiz on the page (low in the scroll) ... the strongest capture device on either build. Greg offers the soft CTA from the hero and nav but links out. The final should do both.

Visual explanation

Greg

9.0

Oliver

7.5

Craft and explanation scored separately. Oliver's hero is the best-looking screen on either build; Greg's hero diagram teaches the entire model in one glance. Different jobs ... the final needs both.

MMF fidelity (V4)

Greg

9.5

Oliver

5.5

The framework everyone signed off on is the judge here, not taste. The widest gap on the board, and the most fixable: the missing pieces are words, not rebuilds. Full rubric below.

PitchKitchen's proposal: build the composite. Your answers refine it.

Take Greg's narrative structure (the diagram-led hero, the spine, the committee, the honest labels) and transplant in four of Oliver's best organs: the security section, the embedded quiz, the five vendor questions, and the evidence section's placement. Section tally: Greg's build leads 8 rounds, Oliver's leads 5, 2 are even. Every contested call carries its decision question inline below ... agree or overrule as you scroll.

Round by Round

Fifteen sections, head to head.

Each round shows what both builds do and what the final page should take from each. Verdict chips: Keep goes into the composite as-is, Graft gets transplanted from one build into the other, Compress earns a smaller footprint. Contested rounds end with a YOUR CALL box ... click your answer right there, with the comparison still in front of you. Every click saves to the shared board, and teammates' dots appear on a question after you've answered it yourself (first reads stay honest that way).

Navigation

Greg's edgeKeep + graft search

Oliver

Dark minimal nav: Platform, Results, About, Careers, one demo CTA, and a ⌘K search. Clean, with a slick search affordance. Careers pulls a buyer off the path, and there's one CTA speed: demo or nothing.

Greg

The agreed sitemap IA (Why Scribe-X, Solutions, Resources, Company) plus a two-speed CTA pair: soft (Practice Health Score) and hard (Schedule a Demo).

The call: not every visitor is demo-ready. The dual CTA catches both, so the final keeps Greg's nav and grafts Oliver's search. Careers belongs in the footer.

Hero

SplitMerge both · decide below

Oliver

"More than an AI medical scribe. Clinical intelligence for the whole visit." Cinematic photography, gradient accent, real client logos immediately below ... the best-looking single screen on either build. The trade-off: the visual is atmospheric rather than explanatory, "more than an X" frames Scribe-X relative to the category it's leaving, and the Remote Healthcare Assistant ... the differentiator ... isn't on the first screen.

Greg

Five headline options stacked for this vote, with the CIP toggle-bar diagram beside each: Pulse/Assist/Live as positions on a bar, the eight-stage journey rail, and the axis reading AI ONLY → HUMAN AT THE CENTER. The diagram teaches the whole model before a word of body copy. The five-stack is a review artifact ... one headline needs to win.

The call: the diagram is the hero's anchor. It's the one asset on either build that answers "how is this different from the AI we already have" in five seconds, and it's the framework's category-defining mechanism made visible ... so the proposed hero is built AROUND it: the diagram as centerpiece, the winning headline riding on it, Oliver's photographic craft as the treatment, client logos right below. Both hero decisions are right here.

The Problem Story

Greg's edgeDecide below

Oliver

A cinematic scroll sequence: "AI automated the note. It didn't fix the job. So the work comes home," resolving into the work-of-one-visit grid. Genuinely memorable when it plays. The trade-offs: it spends four screen-heights on one idea, quick scanners scroll past it mid-animation, it's hidden on mobile, and scroll-driven motion is the page's biggest accessibility exposure.

Greg

Three exec quotes in the buyers' own voices (CMO: "my best docs are charting past midnight" · CMIO: "half my docs quit using it" · CIO: "I can't sign off on offshore data"), then the same problem diagram, then a 7-row AI-only vs AI+human comparison table.

The call: the exec quotes make all three approvers see themselves in the first 90 seconds, and the comparison table is the most quotable block on either page for AI search. If the team loves the cinematic grid, compress it to one screen inside the narrative structure.

Third-Party Evidence

Oliver's edgeGraft into the final

Oliver

A dedicated "Don't take our word for it" section right after the problem: 80%+ tried ambient AI, ~10% documentation-time savings, 32% sustained usage, JAMA cited, framed as "not our numbers, and exactly why the platform pairs AI with human intelligence." The section header ... "The industry already ran the AI-only experiment" ... is the best line on either page.

Greg

The same JAMA numbers exist but sit inside The Proof section, far down the page, after the solution.

The call: placement. Independent evidence lands hardest right after the problem claim and before the pitch ... it turns the page's thesis from opinion into record. The final adopts Oliver's section, in Oliver's position.

Platform Tiers (Pulse / Assist / Live)

Greg's edgeKeep depth + graft calculator · decide below

Oliver

Three clean cards on an AI-to-human spectrum plus the mix-builder: add your providers, see the mix, with the honest note that results attach to each provider's solution and are never averaged. Real interactivity, light cognitive load. Two gaps: no "who it fits" guidance, and Pulse reads as available today (the framework holds it as COMING until it's operationally ready).

Greg

The full toggle-bar treatment: who each tier fits, what the clinic keeps at each setting ("you keep: the most risk and manual effort"), same-day review specifics, MOST PRACTICES START HERE on Assist, and honest COMING / AVAILABLE NOW labels throughout.

The call: "what you keep" prices the risk of buying less human ... the sharpest selling idea in the tier section. The final keeps Greg's depth, grafts Oliver's mix-builder inside it, and labels Pulse COMING per the framework.

Whole-Journey Coverage

Greg's edgeKeep the labels

Oliver

Three intelligence layers (Pre-Visit, Encounter, Revenue + Quality) with tight bullets. Digestible and confident.

Greg

Seven journey layers, each labeled with its real status: LIVE TODAY, REPORTING LIVE TODAY, COMING SOON, ADD-ON.

The call: the buyer this page must win is a CMIO who's watched tools dazzle in the demo and die in the exam room. Status labels are trust currency to that person, and they pre-answer vendor question #4 below (what's live vs roadmap, in writing). Oliver's 3-layer grouping is a good compression if seven feels heavy ... keep the labels either way.

Security & IT

Oliver's edgeBiggest single graft · decide below

Oliver

A dedicated section: HIPAA workflows, encryption specifics (DTLS/SRTP, TLS over 443, at rest), role-based access, audit logging, no EHR integration required, AWS-hosted, BAA-every-time badges. Speaks fluent CIO.

Greg

Security lives in the FAQ and the CIO persona card ... there's no scannable security block in the scroll.

The call: Greg's build names the CIO as a buyer and then makes them dig for their answers; Oliver's section IS the CIO's card, executed. Clearest gap in Greg's build, easiest high-value transplant.

Competitive Alternatives

Oliver's edgeGraft the 5 questions · decide below

Oliver

Four alternatives each framed THE APPEAL / THE CATCH ... honest enough to be disarming. Then the standout: "Questions worth asking any documentation vendor," five questions Scribe-X answers in writing, with "we'd encourage you to ask everyone else too."

Greg

The same four alternatives, named honestly, each with where it stops, closing with "Scribe-X closes all four gaps." Solid and complete.

The call: the five-questions device arms the internal champion for every competing vendor call and sets criteria only Scribe-X can win. The best content idea on Oliver's build ... the final adopts it.

Outcomes / Proof

EvenBoth need verified numbers

Oliver

Outcome ranges sourced to the 2025 client cohort, with a discipline gem: results attach to each provider's solution, never averaged across the platform. One technical fix needed: the count-up animation serves its starting zeros to crawlers, so AI engines reading the raw page see "0.0–3.8 more patients" and "0–62% faster close."

Greg

The same stat family; the raw HTML serves true values (crawler-safe), and the block is stamped "pending verification before publication" ... honest, and obviously not shippable until the numbers are verified.

The call: even round. The final needs the 2025 cohort numbers verified once, rendered server-side as real values, and scoped per solution the way Oliver's framing already does.

Testimonials

Greg's edgeTighten

Oliver

One quote (Dr. MacDonald, Hometown Health): "I walk out of here and my charts are done." Punchy and human.

Greg

Two quotes including a Director of Operations at Adventist Health, framed as "Not pilots. Community health centers that have run the whole-journey model with us for years."

The call: "for years" plus a recognizable health-system name is the anti-churn proof an FQHC wants ... but the long quote needs cutting to its best two sentences. Ideal final state: three tight quotes, one per buyer persona.

Who This Is For (Buying Committee)

Only in Greg'sKeep · decide below

Oliver

No equivalent section ... the committee is implied, never named.

Greg

"Built for community health, where the stakes are highest" with three persona cards ... Champion CMO, Bridge CMIO, Risk-and-Governance CIO ... each with an inner-monologue quote.

The call: enterprise community health is a committee sale. When the CMO forwards the link, the CIO finds their own card waiting. No other section does deal-velocity work like that.

Lead Capture / Front Door

Oliver's edgeGraft the embed · decide below

Oliver

The Practice Health Score quiz is embedded in the page ... question 1 of 4 with clickable answers, sitting low in the scroll just above the FAQ, so a visitor who reads that far starts answering without leaving. The lowest-friction capture device on either build. Note the hero doesn't offer it: the hero pair is Request a Demo + Explore the platform, so the soft offer appears only deep in the page.

Greg

A dedicated front-door section pitching the Health Score, offered from the hero and nav ... but as a button to a separate page. Also shows a "Start a free trial" button, an offer that isn't defined anywhere yet ... its decision sits in this round.

The call: starting the quiz in-page is the difference between "interesting" and "in the funnel." The final offers the Health Score from the hero (Greg's architecture) AND embeds the quiz in the front-door section (Oliver's mechanic).

Cost of Standing Still

Only in Greg'sKeep · decide below

Oliver

No equivalent ... the page never prices inaction.

Greg

Six compounding costs (losing your best people, revenue leaks, dropped patient loops, the unused pilot you're still funding, growing liability, mounting pressure), closing with "the question was never whether to change. It's whether to fix it while it's still cheap."

The call: loss aversion is the strongest lever on a risk-averse healthcare buyer, and "you're still funding an unused pilot" speaks directly to the clinics that already tried ambient AI. This section does the urgency work every demo request needs.

FAQ

SplitMerge the sets · decide below

Oliver

Ten crisp answers styled as a chat ("Ask the platform," answers avatared "AI"). The content is excellent ... FAQ 08 carries the true outcome ranges. One question for the team: does an AI chatbot answering the questions sit comfortably under a brand whose whole story is human accountability?

Greg

Fourteen questions with topic filters, phrased in the buyer's voice ("Our EHR already gives us ambient AI, close to free"), naming DAX Copilot and Epic head-on, covering SOC 2.

The call: buyer-voiced questions that name competitors match real search queries (AI-search gold) and real committee objections. Proposed final: Greg's question set, Oliver's answer tightness, standard accordion presentation.

Final CTA & Footer

EvenTake both

Oliver

"Ready to cover the whole visit?" with demo + Health Score, clean footer with a Security page link (a nice CIO touch).

Greg

"PATIENTS, NOT PAPERWORK. Give your providers their day back." The credo close that bookends the narrative, same dual CTA, fuller resources footer.

The call: Greg's close has the soul (it's the locked credo), Oliver's footer has the utility. The final takes both.

The Framework Check

MMF fidelity: scoring both builds against the framework we all locked.

This is the lens that keeps the whole comparison objective. The rubric is the Magnetic Messaging Framework everyone adjudicated and signed off on (V4, July 6): the spine, the toggle-bar mechanism, the naming decisions, the buying committee, the credo, and the language library. The audit cuts both ways ... each build gets caught once.

MMF element (V4)	Oliver's build	Greg's build
The spine ... "a human at the center, not at the edge"	Absent The framework's central sentence doesn't appear. The hero frames Scribe-X as "more than an AI medical scribe" ... a different, comparative frame. Words, not a rebuild ... very fixable.	Verbatim Hero Option D is the spine word for word; the subhead names the mechanism (AI carries volume, a human owns judgment).
Category name ... Clinical Intelligence Platform	Present Hero eyebrow and throughout.	Present "THE SOLUTION · The Clinical Intelligence Platform."
The mechanism ... toggle bar, set per provider	Present "One platform. Set per provider," with the mix-builder and the AI↔human axis.	Present + fuller Who-it-fits and what-you-keep at every setting.
Solution names ... Pulse / Assist / Live	Correct	Correct
Pulse shown grayed / COMING, never promoted	Needs fixing Pulse reads as live. The framework (§10) is explicit: Pulse stays grayed until operationally ready, and Assist is the start-here tier.	Correct Pulse labeled COMING; MOST PRACTICES START HERE sits on Assist, per §11.
People term ... "Remote Healthcare Assistant (RHA)"	Correct Used in body and meta from the start.	Fixed July 16 The heroes briefly carried the retired term "clinical specialist"; all five now read "US-based Remote Healthcare Assistant." The framework audits everyone ... including PitchKitchen.
The buying committee ... CMO / CMIO / CIO (§6)	Absent Implied, never named.	Full Three persona cards matching the three clinical doors.
The credo ... "Patients, Not Paperwork"	Absent	Present Page title, hero option B, final CTA.
The Shift ... AI-only vs AI + Human Intelligence (§4)	Strong "The industry already ran the AI-only experiment" ... the best framing of the shift on either page.	Strong The 7-row comparison table maps almost 1:1 to the framework's shift table.
The villain ... "Set-and-Forget AI"	Implied Never named.	Implied Never named. Room for both to sharpen in the final.
The whole patient journey ... before / during / after	Present Three layers.	Present Seven stages with status labels.
Cost of doing nothing (§9)	Absent	Present Six compounding costs.

Why this lens matters most for the decision

Three of the framework's load-bearing pillars ... the spine sentence, the buying committee, and the credo ... currently live on only one build. Those aren't finishing touches; they're the framework's answer to "who are we, who buys, and what do we stand for." The encouraging part: every gap on both sides is a copy-and-sections fix, not a rebuild. The composite closes all of them at once.

The Visual Question

Two kinds of "good visual" ... the final needs both.

Oliver's hero is the best-crafted screen on either build: cinematic photography, confident type, immediate trust logos. Greg's hero carries a different kind of asset ... an information graphic that teaches the entire model in one glance. Here's that diagram, recreated:

Clinical Intelligence Platform

In the live hero the bar is a light-to-dark gradient and the handle animates in.

What the diagram accomplishes in one glance

The category name (CIP) · the three solutions as positions on a bar · their availability (Pulse grayed + COMING, Assist and Live live) · the per-provider toggle (the handle) · the whole eight-stage journey · the wedge (only THE NOTE lights up ... the one stage AI-only covers, with seven others untouched) · and the spine itself as the axis: AI only on the left, human at the center on the right. That's MMF §10, the category-defining asset, as a picture. A CMIO asking "how is this different from the AI we already have" gets the answer in five seconds.

Its honest limit: it's dense, and at mobile width the journey labels shrink into the contrast issues flagged in the appendix. The final should carry a simplified mobile version. Oliver's photographic craft applied to this diagram's treatment is the best-of-both hero ... which is exactly what the proposal recommends.

The Live Board

Your calls save as you click. The board updates itself.

Pick your name in the bar at the top, then answer the YOUR CALL boxes as you scroll ... every click saves instantly to the shared board. Teammates' votes show up as colored dots next to each option, and to keep first reads honest, a question's dots stay hidden until you've answered it yourself. Answer all 15 and you've earned the confetti. Change your mind anytime ... your dot just moves.

Anything the questions didn't capture?

Backup: copy my calls to the clipboard if you'd rather paste them to Greg directly.

The Proposal

PitchKitchen's proposed final ... your answers refine it.

1. Hero ... the CIP toggle-bar diagram as the centerpiece, the winning headline riding on it, Oliver's photographic polish on the treatment, trust logos immediately below.

2. Problem ... the three committee quotes + the 7-row comparison table; the cinematic grid compressed to one screen if kept.

3. Evidence ... Oliver's "The industry already ran the AI-only experiment" section, in that position.

4. Platform ... the full toggle-bar depth (who it fits / what you keep / honest status labels) with Oliver's mix-builder grafted in; Pulse labeled COMING.

5. Security & IT ... Oliver's section, transplanted whole.

6. Alternatives ... the appeal/catch framing + the five vendor questions.

7. Who this is for ... the three persona cards.

8. Proof ... verified 2025 cohort ranges, scoped per solution, real values server-rendered. Three tight testimonials.

9. Front door ... the Health Score offered from the hero AND the quiz embedded in the section.

10. Cost of standing still ... the six compounding costs.

11. FAQ ... the merged question set, buyer-voiced, accordion presentation.

12. Close ... "Patients, not paperwork. Give your providers their day back." Dual CTA.

Appendix

The technical receipts.

The greyscale-logo question, settled by the standard

The question came up whether the nav logo should be greyscale for ADA reasons. WCAG ... the standard ADA web claims are measured against ... is explicit: text that is part of a logo or brand name has no contrast requirement (Success Criterion 1.4.3, and the logo exemption carries into non-text contrast, 1.4.11). A full-color logo and a greyscale logo are equally compliant, so this one is purely a design-taste call.

Where the real accessibility work lives, on both builds: body-text contrast (Oliver's amber stat color measures 1.76:1 and the slate labels 2.45:1 against white; Greg's light-blue labels measure 1.96:1 and the amber COMING badge 1.76:1 ... AA requires 4.5:1 for normal text), a skip link (missing on both), a main landmark on Greg's build, and a prefers-reduced-motion fallback for Oliver's scroll-driven section. All fixable in a day on the final build.

Check	Oliver's build	Greg's build
H1 discipline	1 H1, clean hierarchy	5 H1s (the stacked heroes ... resolves to 1 after the vote)
Meta description	Present, strong	Missing (draft state)
JSON-LD structured data	None	None ... the final should ship FAQPage + Organization schema for AI search
Image alt text	31 images, all described	22 images, 10 decorative-empty (acceptable)
Contrast fails (WCAG AA)	11	13
Stats served to crawlers	Count-up zeros ("0–62%") ... fix by server-rendering real values	Real values
Content without JavaScript	Text renders	50 scroll-reveal blocks start invisible ... needs a no-JS fallback
Server-rendered text volume	~12.7K chars	~21.7K chars

Housekeeping items for whoever builds the final

Pulse labeling: whatever ships must show Pulse as COMING (grayed), with Assist as the start-here tier, per the framework.
"Start a free trial": currently on Greg's build with no defined offer behind it. Define it or pull it before launch (the decision sits in the Lead Capture round).
Access gate on the CIP preview: heads-up for Oliver ... Cloudflare Access on scribex-cip-homepage.pages.dev doesn't cover the per-deployment hash URLs (e.g. de5aab9b.scribex-cip-homepage.pages.dev serves without login, which is how this audit read the page). One settings change ... including *.scribex-cip-homepage.pages.dev in the Access application ... closes it.
Language consistency: the July 6 naming decision (Remote Healthcare Assistant, not "clinical specialist"; solution names, not tier names) is now consistent on both builds ... keep it that way in the final.
Before launch, both builds' shared list: verified outcome numbers, JSON-LD, contrast fixes, skip links, and the standard Scribe-X tracking block (GTM + HubSpot).