Three structural directions for the homepage, low-fidelity / greyscale. The site is becoming the internal "OS dashboard" for the offplan.online business — the window into what's being worked on, what shipped, and the state of play. Pick one; a follow-up builds it. This is layout + IA exploration only — not the production visual register.
The homepage is the timeline. "What's New" releases pinned at top, then a single reverse-chron stream of everything — plans ratified, workstreams advanced, decisions made, releases cut. Filters + the artefact library are a secondary view you toggle into. The "newspaper of the business."
The homepage is a control panel. "What's New" releases at top, then a status board — active workstreams by priority, blocked items, what's in-flight — then the timeline as one panel among several, with the artefact library as a tab. The "mission control" view.
kind=workstream, status=active. "See all workstreams →" goes to the full Library tab.The artefact library stays central (it's the substance) but gains a strong default "Timeline" view mode, a clean "Sessions" tab, the "What's New" widget up top, and a prominent Changelog link. The "well-organised filing cabinet that also shows you recent activity." Closest to today's site — lowest-risk evolution.
All three satisfy the seven requirements. The differentiator is risk and reversibility. C is a re-skin of what already works: the filter bar, the group-folded cards, the meta-tag pipeline, the search page — all survive. The only genuinely new pieces are (a) a view-mode switch with Timeline as the default lens over the existing dataset, (b) pulling Sessions out into their own tab, (c) the "What's New" widget (which the releases-system plan already commits to building — R3), and (d) promoting /changelog into the nav. That's a weekend, not a rewrite, and if Timeline-default turns out wrong you flip the default back to Grid with a one-line change.
A (Activity Feed first) is the most opinionated and the most "OS-like," but it's a bigger bet: it demotes the library to a sub-page, which is a real navigation change for the two people who use this daily, and the unified-stream needs a normalised "events" model that doesn't exist yet (today everything is a rendered .md with meta tags, not an event log). Worth revisiting once there's an actual event stream to draw from. B (Dashboard) is attractive — the stat strip genuinely communicates "running a business" — but the status board is only as good as the workstream metadata, and a half-populated mission-control panel looks worse than no panel. Borrow B's 4-up stat strip into C as a slim band under "What's New"; skip the full status board until workstream hygiene is reliably current.
Net: ship C now, with the stat strip. Keep A in your back pocket as the v2 once a real event log exists. The path C → A is smooth (Timeline-as-a-tab becomes Timeline-as-the-homepage); the path A → anything-else is not.
Mostly no — and that's a point in C's favour. The per-document renders (the canonical Skeleton-White visual register with oxidised pills) are good and not in scope here. The search page (/search/) is fine as-is. Two things do want attention, and both are already on the books: (1) the session about-pages — the releases-system plan's R7 already kills them so a session click lands on the full doc; that should land regardless of which homepage variant wins. (2) The per-artefact about-pages for non-sessions are a slightly awkward middle layer between the index and the full render — worth a separate look at whether they earn their keep, but not blocking. Everything else (the deploy pipeline, the meta-tag contract, the renderers) is sound; the homepage is genuinely the only structural gap, which is why this exploration is narrowly scoped to it.
Wireframe file: docs/rendered/_wireframes/homepage-redesign.html → after CF Pages deploys, https://preview.offplan.online/_wireframes/homepage-redesign.html. Greyscale / low-fidelity by design — production visual register is the existing Skeleton-White / Helvetica Neue / oxidised-pill system, not explored here.