Title + the shell period pill (landings obey it)

One toned sentence, at most one action

One door per child listing: count as a label, one judgment

Every team at once through one lens — area = people, fill = the lens; lens tabs incl. Harness ▾; hover a tile for its people

One c67 card per team, worst first, with the lens's micro-form (segments · glyphs · meter)

Team × the five /team lenses, shaded per column; the active lens sorts the rows; click a header to switch

Is everyone covered, who needs me? Team blocks in calm grape, the ≤5 who need you ringed and named; hover joins them

One block per team, every person toned by their own state, trouble in the same corner of each block

One stack per team: the calm mass sits low, gaps and trouble make the ragged top

Every lens at once, one DataMachine card each (AI usage · Spend · Active · Violations · Coverage · Harness · Waste · Retention) in three columns — the top six teams named and drawn, the rest counted

Three cards — Adoption (AI usage · Active · Retention · Harness), Cost (Spend · Waste), Risk (Violations · Coverage) — each title a lens select scoped to its group

What is our AI made of, and how concentrated? variant = alluvial · switchboard · bowtie · body; lens = tokens · spend · installs (shown here on installs)

Harness → provider → model as one image; unseen installs as dashed stubs, personal accounts hatched

A dot per harness–provider pair, rings for personal accounts; the scale answer

One row per provider: harnesses routing in, models it serves

The census, one pixel per asset: head models · arms MCP / skills / tools · torso installs · hips workspaces · feet providers

What can’t we see? Monitored installs as a block, the detected-only pulled out and named

Declared, seen, or both? MCP servers, skills and tools by witness, hollow = unvetted

What got in this window? One dot per first-seen asset by class; the sweep on first run

Whose account is it? Named standout for 1–4 personal accounts; per-provider org · dual-use · personal · unknown

The first-run reveal: what the sweep found class by class, and how many installs report

The same room re-lit for one reader, in the URL (?as=) — each specialist carries how many things would wait for them

First run: what discovery has found so far, and the two steps that wait on the customer

≤5 rows for one reader, one verb each — a projection of Findings, decisions and hygiene

Every install, a room per team, toned by the lens — portal /fleet

One pixel per install, a block per team — the pixel-field grammar, here seen as IT

Harness → provider bowtie, toned by the lens — portal to the Inventory mock

Team tiles sized by people, toned by the lens; team hover breaks into people — portal to the Team mock

One field, a person per mark, teams as contiguous runs — here seen as Engineering

The live pulse over open findings on the remediation clock — portal /findings · /live

The meter wall: frameworks and policy pips, or the lens's own goals — portal /controls

The evidence shelf with freshness, and the integration lamps — portal /integrations

Security / Compliance: enforced actions paired with denials — neither side alone

Finance: a bar per campaign — spent, labelled 7-day projection, ceiling

IT: named machines, one verb each; the rows re-add to the coverage gap

Engineering: joiners against their team's p25–p75 band; stalled → check in

Seven days ending today, every event on its day; a lens keeps its own lit

Can we believe this page? Coverage, sensors, capture, evidence — promoted when the eyes are broken