A sharper console, a living landing page, and cards that say why

What changed for people using Bellwether

• The landing page moves. A field of fine particles drifts behind the opening line and settles around the words; a small chevron under the headline says "scroll"; further down, the desk's daily loop draws itself as it completes a lap. On a phone the page fits the screen and every link is thumb-sized.
• The console is easier to read. Type sizes, spacing and contrast were re-measured on every screen: titles look like titles, muted text is readable, numbers are never cut short, and a label stays with its value. On wide screens a quiet column of particles lives in the empty margins, never behind a number or a table.
• Approval cards say what and why in fields. A card that needs your decision leads with one sentence, then a short row of figures (target size, what you hold, the stop), then the signal in words instead of the desk's raw reasoning. Core index-fund top-ups say they are part of your core; a protective exit says so.
• "Bellwether's mind" reads as fields. Each name the desk is watching is one line: what it would do, at what price, and what holds it back. The Critic's objections are grouped per name with the point one click away. When you change posture or autonomy, the card says your plan predates the change and when it re-plans.
• The activity log reads as sentences. "Bought 12 NVDA at $128.55", "Posture changed to Speculative", "The broker's protective stop sold AAPL", with codes and ids one click away under "Technical detail". A link from a chart marker that cannot find its row says so and offers to look in older rows.

What you may notice

• The left rail's group headings (Monitor, Shape, Explore, Manage) are brighter and separated by a short rule; the sign-in and set-up screens sit on a calm particle field instead of the old ember glow.
• Overview alerts speak plainly ("An order for NVDA was not filled before its 14:30 CT deadline") and say whether they need you; the live-prices chip explains an interruption in words and says when prices will refresh.
• On a symbol's page the four things you can do are spelled out: put it on the watchlist, request research, ask the desk, or place an order yourself. Only the last one trades.
• Position rows show a small weight bar against your book's real cap with the words beside it, and a thesis chip that says "written today" or "12 days old" instead of "0d".
• The step-up dialog (your authenticator code before a sensitive action) is one short instruction and one input; the People sheet names every book of yours someone is not in.
• Markets tiles, the Quant card, the Autonomy tuning block and the desk's role cards were re-worded and re-spaced.
• Help screenshots were re-shot to match the new look.

Known gaps and what’s next

• The typography clean-up of page introductions and the help drawer's tables follows in the next console release.
• A richer particle engine (denser fields, formations that mean something, a mark for "an agent is working") is being built to replace today's fields.
• A few mind-card sentences still open with a label and a colon; they are being reworded.