Every option below starts from the same fixed ingredients: your two-sentence intro, the shipped keynote theme (espresso / paper / orange, Instrument Serif), and the six live renderers. What varies is the layout thesis — what the first viewport claims the page is for. Each mockup is live HTML in the real theme with real chart screenshots, not a sketch.
FIXED IN ALL FIVE — intro copy: “safety.viz is a charting library for monitoring clinical trial safety. It’s an agent-assisted update of the safetyGraphics framework.” (keynote + safetyGraphics links) · shipped as the baseline in PR #30
Say less; change nothing else.
The structure you have today, with the lead block cut to the two-sentence intro (this is exactly what PR #30 ships).
What changesGain: zero risk, already built. Cost: the first viewport is still mostly furniture — you read about charts before you see one.
Choose this if You want the copy fix now and the layout conversation separately.
safety.viz is a charting library for monitoring clinical trial safety. It’s an agent-assisted update of the safetyGraphics framework.
Prove it’s real in the first viewport.
Split hero: intro and CTAs on the left, a full rendered histogram on the right. The galleries merge into one grid with a one-line queue strip.
What changesGain: a chart above the fold; the page leads with evidence. Cost: the hero image needs to stay current with the renderer (already automated via the heroAsset hook).
Choose this if You want a conventional, confident product page.
safety.viz is a charting library for monitoring clinical trial safety. It’s an agent-assisted update of the safetyGraphics framework.
Queued: Paneled Outlier Explorer · Adverse Event Explorer · Web Codebook — requirement matrices reviewed
The demo is the homepage.
A slim intro bar, then the actual interactive histogram — real sidebar, real data, running from the committed bundle — before any prose. Gallery below as a compact row.
What changesGain: a visitor is interacting with clinical safety software in under five seconds — the strongest possible claim. Cost: the most engineering (demo lifecycle on the gallery page, data payload on first load); needs care on mobile.
Choose this if You want the R/Pharma audience to touch it immediately.
Six charts, pick one.
The intro sentence, then straight into six large renderer cards. The migration queue compresses from three placeholder cards to one mono status line.
What changesGain: fastest route to content; the homepage is a catalog, which is what a charting library is. Cost: no single focal point; relies on thumbnail quality (real screenshots — already in hand).
Choose this if You believe the six real charts are the argument, and prose is overhead.
safety.viz is a charting library for monitoring clinical trial safety. It’s an agent-assisted update of the safetyGraphics framework.
In the queue: Paneled Outlier Explorer · Adverse Event Explorer · Web Codebook — each already has a reviewed requirement matrix
The exhibit page.
Your two sentences set huge in the display serif — the intro is the hero. One orange CTA, a filmstrip of the six renderers, the queue as a footnote.
What changesGain: unmistakably yours — it reads like the keynote, memorable for the talk. Cost: least conventional for a tool homepage; the filmstrip buries the per-renderer links one click deeper.
Choose this if You want the site itself to be an artifact of the talk.
safety.viz is a charting library for monitoring clinical trial safety. It’s an agent-assisted update of the safetyGraphics framework.
Six renderers live · three in the queue · every one requirement-traced
D (Gallery first) matches how you’ve steered this project all day: real charts over prose, real data over placeholders, show-don’t-tell. A charting library’s homepage that is the chart catalog needs no further explanation, and the two-sentence intro is exactly enough framing. It also degrades gracefully — it’s the baseline plus bigger cards and a demoted queue.
B (Chart forward) is the runner-up if you want one hero moment for first-time visitors (and screenshots of the site for the talk) — it keeps a focal image without adding prose back.
C (Demo-first) is the most impressive and the most work; it would make a strong v1.1 flourish on top of D (swap the top card row for the live mount once it’s proven on mobile). E is worth keeping in the drawer for a dedicated talk-companion page rather than the product homepage.
Next step
Pick a letter (or a hybrid — “D with B’s hero” is coherent) and I’ll implement it on the PR #30 branch with tests and preview.