Configured workspace
What should we look at?
Data as of 08:10 · 4 sources
Suggested
Recent
CelesnityConfigured
MinderMinder · AI chat
AI chat
Threads on the left, the conversation in the middle, and a stage of panes on the right that opens only when there is something to show.
Overview
The whole surface, live. Send a message, open a card, split a pane, switch threads.
Loading the workspace…
- Threads rail
- 264px · collapses to 52px · becomes a title menu below 1040px
- Chat
- min 340px · reading column 768px · 44% of the split by default
- Stage
- min 380px · 56% · appears with the first artifact, never before
- Below 760px
- the stage stacks under the chat; on a phone it is a sheet
- Splits
- right adds a column, down stacks · 2 × 2 at most · a new kind opens a new pane, never a modal
Starting
Before the first turn: who you are talking to, how fresh the data is, and where to begin.
First run
Let’s give Minder something to read.
Nothing is connected yet.
Start with
CelesnityNothing connected
- Heading
- 26px / 600 · one line under it: data as of, number of sources
- Starters
- 3 suggested rows (36px) · 2 recent, only when there is no rail · 1 signal worth acting on
- Composer
- at the bottom of the empty state, where it stays for the conversation
Composer
Chips say what the agent can see. The bar carries attachments, mode, effort, dictation and send. Type / for tools.
Composer
composed · input-group + dropdown-menu + popover + kbdCelesnityConfiguredorders_dailyv12shift-plan.xlsx48 KB
↵ send⇧↵ new line/ tools@ a table
- Box
- radius 16 · field 44–180px, 16px text · focus darkens the border, no ring
- Chips
- context 26px pills above the text · toolbar chips 28px
- Bar, in order
- + menu · mode · effort · dictate · send (Stop while the agent works)
- Narrow
- under 460px the mode chip keeps its icon only; under 380px dictation goes · Send never leaves the box
- After sending
- the field drops back to one line
Messages
The person is a bubble; the answer is not. One reading column for both, and four things an answer can carry.
A question and its answers
One run: orders_daily stopped at 02:14 on a mapping error. Nothing downstream of it has refreshed since.
Six orders are past their promised date. Five sit behind the failed run — their picks were never released, and they clear as soon as it is repaired. The sixth is with the carrier and unrelated.
- Five held by the run — clear on repair
- One with the carrier — needs a revised date
Queued. I will re-run the window once the mapping is corrected and release the five.
What an answer can carry · link · file · source · output
Six orders are past their promised date. I read the pipeline's own tables rather than the reporting copy, which is a day behind.
02:14:07 INFO window 2026-09-08 → 2026-09-09 02:14:07 INFO reading 41,882 rows 02:14:11 WARN carrier_ref: 6 unmatched keys 02:14:11 ERROR mapping 'carrier_ref' → null for 6 rows 02:14:11 ERROR aborting before write 02:14:12 INFO 0 rows committed 02:14:12 INFO downstream views not refreshed 02:14:12 FATAL run failed after 5.2s … truncated
The six unmatched keys are the six late orders. I have put the row-level detail and the mapping as it stands next to each other:
The carrier's own tracking disagrees with our reference on two of them:
A long answer · headings and bullets inside the turn
Output finished the week at 9,140 units, 4% under plan. Two things account for almost all of the gap, and only one of them is worth acting on.
Where the units went
Line 2 ran all seven days and carried 41% of the volume, which is the most it has taken in a quarter. Line 3 lost most of Tuesday's afternoon shift to a changeover that ran roughly ninety minutes long, and never made the time back — the following days were already committed.
- Line 2 — 3,747 units, no stoppages
- Line 3 — 2,180 units, one long changeover
- Line 1 — 3,213 units, one quality hold
What is worth acting on
The changeover is the real loss and it is repeatable: the same product pair has run long three times this month. The quality hold on Line 1 was a single 40-unit batch, already quarantined, with no pattern across the week — it reads worse in the totals than it is.
If the changeover sequence is corrected, last week's shortfall does not recur. Nothing else in the week suggests a second cause.
Arriving
Reduced motion turns this off rather than shortening it.
One run: orders_daily stopped at 02:14 on a mapping error. Nothing downstream of it has refreshed since.
Six orders are past their promised date. Five sit behind the failed run — their picks were never released, and they clear as soon as it is repaired. The sixth is with the carrier and unrelated.
- Five held by the run — clear on repair
- One with the carrier — needs a revised date
- Question
- secondary bubble, right-aligned · radius 18 · padding 10 × 16
- Answer
- no bubble, no avatar, no name · 16px / 1.65 · column 768px
- Link
- the host beside it and ↗
- File
- an attachment chip; the person's own can be removed
- Source
- the dataset in mono · a filled dot when live, a ring when bound
- Output
- mono, capped at 9.5rem, scrolls inside
- Arrival
- 0.25s ease-out, 4px rise · none under reduced motion
A turn
From Send to a finished answer: a status that names the running step, words that arrive as words, a chart where it is announced, and a footer that is already there.
Working · nothing to name yet
Thinking…
Working · a step is running
Thinking…Reading table structure · iceberg.factory_silver.press_telemetry
Listing tables · iceberg.factory_silver
Working · a step failed
Thinking…Drawing the chart
Listing tables · iceberg.factory_silverRunning query · hourly averages
Working · under text already on screen
Line 2 carried 41% of the week's output. Let me check what happened on Tuesday —
Thinking…Running query · production_runs, 1–7 Sep
Words as they arrive
Chart · while it is drawn
Here is the hourly average for the most recent day in the table:
Chart · landed
Here is the hourly average for the most recent day in the table:
Average press temperature per hourpress_telemetry · 2 Sep
Temperature climbs through the morning and peaks at 66.3 °C at noon, above the usual band.
The newest answer
Run #418 failed at 02:14 on the transform step — the source table gained a column the mapping did not expect.
Ask more about this
09:14
An older answer · hover to see its actions
Six orders are late. Five sit behind the failed run; the sixth is genuinely stuck at the carrier.
09:12
- Status card
- neutral tint, radius 12, 36px orb tile · “Thinking…” 15px / 600 · caption 13px: the step, its target, seconds · role=status
- Nothing to name
- “Thinking…” alone · no rotating guesses
- Steps
- finished ones fold into chips under the card · done is grey, failed is red · none once text is on screen
- Words
- a word per tick (~30/s), each fading in over 450ms · a half-written word, unclosed ** ` [ and agent tags are held back
- Chart
- after the sentence that announced it · 68px header, chart / table toggle, open in side panel · a skeleton of the same size while it draws · its stage tab opens at once
- Tables, documents
- stay one-line cards that open a pane
- Follow-ups
- under the newest answer · at most 3 rows, 16px, hairlines · “Ask more about this” 18px / 600, left out when the prose ends on its own lead-in · live only once the answer ends
- Actions
- copy · regenerate (newest only) · time · run details · 32px ghost buttons · older answers show them on hover and keep the row's height · nothing disabled for later
- Scroll
- sending brings the question to the top once · nothing follows the stream · ↓ appears when the answer outgrows the view
- Motion
- all optional · under reduced motion nothing fades, glides or breathes; the order stays the same
Stage
Every pane wears the same header. Tabs sit in a numbered row beneath it once there are two.
Pane header and tabs
composed · button + dropdown-menu + tooltipWeekly operations brief
v3/3
The stage's outer edge · one pane · split · a ground pane
Weekly operations brief
Output
Late orders
Board
Late orders
Generating
Output per line
Querying production_runs · 2 of 3 steps
Failed
Late orders
Could not read orders
The connection to ops-db timed out after 30 s. Nothing was changed.
Behind the latest version
Weekly operations brief
v3/3
You are reading v2. The brief was regenerated 4 minutes ago.
- Header
- 38px · the active tab's title left · versions, share, more, open out, maximise, close right, in that order
- Tabs
- a numbered row under the header from two tabs · one kind of thing per pane · pressing one scrolls the transcript to its card
- Which card is open
- the one whose tab is in front, not every card with a tab
- Frame
- pane radius 10 · 8px gutters on a tinted ground · a single pane goes flush
- States
- generating: marker + skeleton · failed: alert + retry · behind: banner + version pager
- Compact (under 480px)
- share and open out fold into the ⋮ menu
Threads
One list, three homes. A dot marks a thread that needs you.
Rail
Rail · collapsed
Switcher · narrow
Thread: Weekly ops reviewOpen the title to switch; ⋯ holds the same actions as the rail.
- Rail
- 264px, collapses to 52px · rows 32px · Pinned, Today, Yesterday, Previous 7 days, Older
- Dots
- 7px · blue running (pulses) · amber waiting for approval · red failed
- Narrow
- the conversation title becomes the switcher · 320 × 420 panel
- Row menu
- pin, rename, share, schedule, archive · replaces the time on hover
Primitives
The components the surfaces above are built from, with their variants.
Message · with the multi-agent slots
Minder AI
Answered in 1.2s · 4 sources
Bubble
Marker
Reading 4 files…
Deployment plan drafted
Today
Context updated
Attachment
Questionnaire · when the agent needs a decision it cannot infer
composed · field + radio-group + progressQuestion 1 of 20%
This shapes how it ranks the next actions.