Updated 26 Oct · Design Status: bookstore events feed proposal (26 Oct); Home Base flow audit (26 Oct)

design status

Updated 23 Oct · Where each surface stands
Proposal only · nothing built

Art Matters · an events feed from bookstores (26 Oct)

Once you've said yes to the shops in the "Bookstores to watch" blocks (New York City, Los Angeles, Paris and London guides), we'd read each one's events page once a week, every Monday, and pick out the events that fit Art Matters: art-book and photobook signings, launches with artists and designers, writers in conversation, film and food tie-ins. We'd leave out celebrity memoirs and book-club nights. The good ones would show on Home Base as "This week, IRL" for you to keep or drop, and the ones you keep would go into Friday's Substack under the city, with the shop, the date and the link. Where a shop only posts on Instagram (Ofr., Donlon, Mast), we'd note that and check by hand. Nothing reaches readers without your yes.

Start here

To market · your design work (25 Oct)

In 30 seconds: the deep work is built: the desks, the records and the nightly research. What stands between it and readers is design, and eight calls only you can make. Each row below has our suggestion filled in. Tap Yes, go with ours and it reaches us as a note (it shows in Your notes on Home Base), or tap Something else to say what you want.

NowDesks, records and nightly research built
→
You · rows 1–5The reader card, disclosure on Proof, the Directory, the library, Parts
→
WeWrite the briefs; clean up cards with your OK
→
You · rows 6–8Media card, search, social and widgets
→
We build, you reviewCard, tile, library and embeds
→
YouVet and Put live on the Brand Desk
→
LaunchReaders see it on canava.co
  1. The reader's brand card open the brief ↓
    Why it blocks market: the library, the tile and the embed all come from this card. No brand reaches a reader until it's decided.
    Photo on a phone: full width, height capped so name, verdict and one line fit the first screen
    Corners: about 5px, one system for all four areas
    Held items: investors when sourced; debate for stories; impact only with a source
    The seven values: show once one is rated, rest dimmed
    Fashion houses: B, link (same as Directory 5)
  2. The disclosure card on the Brand Desk's Proof tab open Fashion Week, then Ruadh ↗
    Why it blocks market: disclosure is the CANAVA signal, and the questions apply to every brand. Today Proof shows the plain six-question list from 10 Oct, so you vet without the card that makes the answers readable, and the reader card's "Made and sourced" has nothing designed to draw from. Design call only; nothing gets built from this row.
    Proof tab: the Fashion Week card (Fibre, Cloth from, Made at, Owned by, the x of 102 bars, the origin quote) replaces the plain list, beside certifications and research; an empty box says To check
  3. The Directory plan open the plan ↓
    Why it blocks market: the Brand Desk holds 4,988 cards and only 941 are brands. Until each card has its right home, research can write brand facts onto a non-brand, and the list you vet from isn't the list readers will see.
    The name: Directory
    Cards with no role: 2,183 get the brand role, logged and undoable; 584 go through cleanup
    Three new roles: add platform, nonprofit and media
    Shops: no own label leaves the Desk; own label sits in both
    Houses: B, brand role once it has a shop
  4. Sign off the brand library open the library ↗
    Why it blocks market: it's the backbone and the front door for brands, open since 3 Aug and empty today. It takes the tile from row 1.
    Sign-off: approve the layout, with row 1's tile
  5. Parts · one link per product open Parts on Buffalo Brussels ↗
    Why it blocks market: recipe pages earn through their links, and today a link can live in three places that disagree. The compact rows are live (14 Oct); this is the one piece of the draft still yours.
    One link field: on the product; the line's own brand link goes
  6. How a media title opens open the live reader ↗
    Why it blocks market: the reader is a placeholder, and a title opens two different ways today. It should open the way a brand will.
    Opening: side sheet on desktop, bottom sheet on a phone
  7. Search readers can use see the gaps ↓
    Why it blocks market: Media Diet readers can't search at all, and canava.co's search can't see anything on the tools site.
    Start: one shared search, Media Diet first; site search stays for articles
  8. Social and widgets, one family see the briefs list ↓
    Why it blocks market: social is how a launch travels. Only Travel has sharing built, and six widgets use four frames.
    Direction: social from Travel's share pattern; widgets to 3:4, one corner
  9. Brand relationships · plan (25 Oct) open the plan ↓
    Why it matters before others join: one private place per brand for when we talked about them and when we talked to them. Written, not built. The first call is about keeping it private.
    Lock first: closed home and team list before any note is typed in
The detail behind each row

Order. Rows 1 to 3 come first because everything a reader sees about a brand depends on them. Row 4 waits on row 1's tile. Row 5 stands alone and can be decided any time. Rows 6 to 8 follow the briefs list further down this page.

Row 2. The card lives on the Fashion Week page (open Ruadh). The Brand Desk's Proof tab today shows the six disclosure questions as a plain list (10 Oct). Our suggestion puts the card beside certifications and research for every brand the questions apply to. Nothing is built until you say.

Row 1. The full reasoning and the ten parts of the card are in the consumer brand card brief (23 Oct). The photo at full width on a phone is about 520px tall, which pushes the name off the first screen. Corners: the August brief and Art Out East are square; Design Rules and the card family use about 5px. Values: as of 20 Sep only 32 of 2,612 brands carried a source.

Row 3. From the Directory plan (24 Oct). Nothing is deleted in the cleanup; every change is logged and can be undone. Decision 5 here and decision 5 in row 1 are the same 20 Sep question, so one answer settles both.

Row 4. The library brief is the 24 Jul draft (not locked). All 12 earlier brands were taken down on 27 Sep for full vetting, so the library shows nothing until you put brands live.

Row 5. The draft is RECIPE-PARTS-DRAFT-2026-10-14 in CA_BUILD › Active. Its compact rows and the Map folded into Parts went live on 14 Oct, on your go (RECIPE-PARTS-LIVE-2026-10-14). The line's own brand link was kept in the opened row until you decide on one link field.

Row 6. The 13 Oct media card design pass is a draft; the live reader uses a side drawer and the beta a tabbed pop-up.

Row 7. The 19 Aug search and filtering brief covers this; its approval isn't recorded.

Row 8. The share and distribution brief and playbook are from 8 Jul. The widgets in use: place, Art Out East, stay, brand, media and recipe.

What a tap does. Yes, go with ours sends one note to us with the row and our suggestion. Something else opens the note box for your own words. Nothing on the tools changes until we act on the note. If you're not signed in on this device, the tools' sign-in box opens once.

Check · 26 Oct

Trawl: what broke (26 Oct)

In 30 seconds: saving still works. All 125 captures in the last two weeks saved, none were lost, and “already in” now adds to the record instead of erroring. What broke is everything after the save: captures stopped being sorted on 25 Oct, the Trawl Log pile has not been worked since 21 Oct, and since 6 Oct most captures go in with no words on them. That is why things end up dropped in chat instead.

WhatStateWhen, and which changeThe fixEffort
Sorting each capture (place, brand, recipe…)Working, checked 26 Oct. It never stopped: it runs every night and sorts 30. On 25 Oct it spent its 30 on old March captures, so the newest waited a day. Last night it sorted every capture since 25 Oct. 133 old March and April captures are left, about 30 a night.
Was: every capture since 25 Oct, midday, is unsorted.
25 Oct. Same days the background jobs started failing: 25 failed runs on 22 Oct, 64 on 25 Oct, 126 today, across 17 jobs, mostly “couldn’t start in time”.Lighten or space the heavy background jobs (the same two slow refreshes as in the Home Base audit), then restart the sorter and run it over the captures it missed.Medium
The Trawl Log pileNot being worked. 1,063 captures sent to “Trawl Log” are waiting; the oldest is from 18 Mar; the last one moved on 21 Oct. 84 arrived this month.Not one change. Nothing moves them on its own, so anything sent to Trawl Log stays there.Move the clear ones on their own (sorted, with a link) and bring the rest to you as one-tap rows, a few a day.Medium to large
Words on a capture (vertical, sub-category, tags, “+ add”)Fixed 26 Oct (once Trawl is published). Words you pick from the shared list now go on the capture’s tags too, and the caption from the link is kept instead of being wiped by an empty “Why”. Of the 49 with no words, 30 now have words from what they are (21 Wellbeing, 5 Media, 2 Beauty, 1 Home, 1 Style); 19 are only a name or a bare link, so nothing to go on.
Was: only 11 of 125 used the list; 49 had no words.
6 Oct, the switch to the shared word list. The old rows are still there underneath; a per-device “use the old lists” switch exists.Bring the free tags box back alongside the shared list, and make the shared list open with your usual picks showing.Small
Place tagging on captureFixed 26 Oct. Cause: Instagram captures were saving with no caption (see Words), so there was nothing to read. Also, short names like OR and ME matched the words “or” and “me”; they now only count in capitals. October re-run: 9 captures got their places (Santa Cruz, Lapland, Puglia, LA, two in New York, London, Tokyo, Georgian Bay), 8 wrong ones taken off. “Olive” (a town in Ulster County) still catches olive oil and names.
Was: 0 to 3 a week since early October.
From the week of 5 Oct. Lines up with the 6 Oct word list change; not yet confirmed as the cause.Check what it reads now that the Note is folded into Why; re-run it over October’s captures.Small
Link from the capture to its home recordFixed 26 Oct: 8 of 10. Linked the 4 recipes, 2 editorial, the Media item (OTO from Tokyo was already on the Media Desk) and the brand post (linked to all six fibre brands). Left for you: the research one (houseofizzie, marked “added by mistake”) and the Build one (Stefan Vignogna’s kitchen details, which reads like Design, not Build).
Was: 10 captures in two weeks with no link.
Mostly 21–22 Oct, the days many captures came in from chat rather than through Trawl.Link the 8; send the 2 on. Have chat filing use the same save as Trawl so the link is always made.Small
“+ Capture with marks” buttonWorks, but it always files to Design reference, whatever the capture is, and sits on top of the page.21 Oct, added as a trial.Keep it only on Design reference, or label it “to Design reference”.Small
Capture (the quick list page)Not used since 1 Jun. It keeps its own list, and when it sends something on it makes no Trawl Log copy and doesn’t check for “already in”.Not changed; left behind as Trawl grew.Retire it and point it at Trawl.Small
Still workingSaving (no failed saves, nothing rescued in 14 days) · “already in” adds to the record (7 Oct) · Instagram and link capture with title and source (59 Instagram captures) · photo upload · every destination: recipes, brands, editorial, resources, media, places, trips · the Trawl Log page itself (one-tap suggestions added 26 Oct). Share sheet, bookmarklet and offline saving were never part of Trawl in any saved copy, so nothing there was lost.

Your calls · our suggestion filled in

  1. Get sorting going again
    Sorting: lighten the background jobs, restart, catch up since 25 Oct
  2. The 1,063 waiting in Trawl Log
    The pile: clear ones move on their own; the rest 10 a day, newest first
  3. Words on a capture
    Words: keep the shared list, bring free tags back
  4. The 10 captures with no link
    Unlinked: link 8, send 2 on, chat uses Trawl’s save
  5. “+ Capture with marks” and the old Capture page
    Marks button: keep, labelled “to Design reference”
    Old Capture page: retire, opens Trawl
Audit · 26 Oct

Home Base flow audit (26 Oct)

Each row on Home Base: where it lands, whether you can tell what it is, whether one tap does it, and whether it leaves Home Base afterwards. Clear hides a row from Home Base only; the record behind it is never removed.

RowLandsUnderstandableOne tapClearsResult
Brands · “Add with via stockists?” and “Load the full range?”Each brand name opens its Brand Desk card; Work through opens them one by oneYesYes: one tap per brand (first tap on a device asks for the 6-digit code once)Yes, the brand drops offWorks
Brands · done rows (e.g. Clare V. queued)Brand Desk cardYesNow: Clear, with UndoNow yes; folded into “Done today (n)”Fixed
Recipes · lines that need your choiceTwo lines shown on Home Base, all in Recipe DeskYes: the line, the question, what each option doesYes (code once per device)Yes, the line drops offFixed (was “Sign in to see them”)
Food · recipes with a To check (new)Each recipe in Recipe DeskYesAnswer or Resolved on the cardRow leaves when the count reaches 0, or ClearNew · Recipe Desk is slow to open (about 15 s)
Food · ingredients without checked nutrition or price (new)Ingredient libraryNo longer on your list (26 Oct). One line in Updates & scans: "Research running · X left", gone when nothing is leftNothing for you to doNoneChanged 26 Oct · kitchen tools no longer counted
Media Diet · titles waiting for your rating (new)Media Desk, Media Diet tab; three names open their cardsYesRate on the cardClear, or it drops when all are ratedNew · the tab shows all 130, with no “not rated” filter
Media Diet · To checks and the “Must read” draft mix (new)The card; the Media Diet tabYesResolved on the card; Put live in Media DeskClearNew
Reference · Media: In or OutEach title now shows cover, one line, Site and Instagram linksYesYes: In or Out (code once per device)YesFixed: only the 68 ready ones are asked; 160 sent to research, 14 people sent to review
Reference · done row (Confirm now routes…)Trawl LogYesClearYesFixed
Build · places to check (Other guides, Millbrook, Hudson, Upstate)Guide Curator for that guideYesNo single tap: each place has its own To checkNow: ClearFixed (had no button at all) · Guide Curator takes 8 s+ to open
To do · Places waiting in the Trawl LogTrawl Log, places onlyYesLink & close, Confirm: place, or Not thisCount drops as each goesFixed: the list showed 177, Home Base said 159; now both count the Trawl Log only
To do · Trawl captures not sorted, Cabagges blocks, Waiting to come in, Maine board, To market, Not a brand?, Rec LeagueThe right tool, on that setYesOne tap per item in eachRow drops when the list is emptyWorks
Updates & scans (new): morning update, briefing, media scans, sales scans, sale eventsEditorial, Notion, Good Shit On Sale, Pipeline HealthYes, each with its last runNothing to tap; a stopped job says so and links to Pipeline HealthNot a to-doNew · Sales email scan is stopped since 24 Oct
System status · last 24 hoursPipeline HealthYesWas “Sent to Dispatch” (untrue); now “Needs a restart — on Pipeline Health”Not a to-doFixed
Affiliate linksBrand Desk, links listsYesLinks to check: Right or Wrong product per linkYesFixed: the read failed when the page opened ~50 reads at once
Affiliate adminClosed by default: “24 to do · 22 done”YesOne tick eachTicked rows move to DoneFixed
Gift listCount, three names, then “all 19 →”YesOpens each cardNot a to-doFixed
Chat backup line (new, top of page)Pipeline Health, fix 4YesNot a to-do; never clearableNot a to-doNew · red: no copy on record since 27 Sep

Holes in other tools, not fixed here

WhereThe holeThe fix
The database, every few minutesTwo background refreshes (the dashboard, and the Brand Desk cache) take 11–12 seconds each and run every few minutes. While they run, pages that open many reads at once (Home Base, Brand Desk) get “ran out of time” answers and show blanks or “busy”.Run them less often or make them lighter. Home Base now spaces its reads and retries, so it fills in a few seconds later instead of blanking.
Sales CuratorOpens on a “curator key” box, a dead end for you.Use the same sign-in as the other desks. Home Base now links to Good Shit On Sale, which works.
Sales email scanStopped since 24 Oct; it had filed 64 sales in two weeks.Restart it (Pipeline Health).
Media researchSwitched off since 4 Oct, so the 160 media titles sent to research wait.Restart it (Pipeline Health); the cloud copy needs its one key.
Media DeskA person filed as media can be found (Home Base link) but not moved to People in one tap; the Media Diet tab has no “not rated” filter.Add “Move to People” on the card, and a Not rated filter on the Media Diet tab.
Design reference (released to another session)Media list needed filters (Ready to decide · Waiting for research · People filed as media) and a one-line description per row. Written into the page source on 26 Oct, not shipped by this session; and no way to edit a card (being built now).Ship with the card Edit work; until then the Media tab shows every title, with no filter.
Recipe DeskOpening a recipe from a link takes about 15 seconds; it reads “Loading the library…” first.Load the one recipe first, the library after.
Guide CuratorOpening any guide shows “Loading…” for 8 seconds or more.Same: load the guide asked for first.
Brand DeskThe links lists sometimes open on “The database is busy… trying again”.It retries on its own; a lighter first read would stop it.
Ingredient libraryNo filter or link for “without checked nutrition” or “without a price”.Add both, so the Food rows can open on just those.
Brand calls on Home BaseThe call title (e.g. “Add with via stockists?”) opens a Notion page that needs a Notion sign-in in Safari.Point the title to the work-through in Brand Desk instead.
Sign-in on SafariEvery save (Yes, In/Out, recipe choice) needs you signed in on that device; the first tap asks for a 6-digit code.Home Base now says so up front with a Sign in now button. Sign in once per device.
New plan · private to the team

Brand relationships · plan (25 Oct)

In 30 seconds: a Relationship tab on each Brand Desk card with two timelines side by side. We talked about them fills itself from the stories, gift lists, Rec League and sale posts we already record. We talked to them holds emails, calls, applications, affiliate status, samples and events, each with who on the team did it. A status line sits on top, and one desk-wide list shows who is waiting on a reply and who is due a follow-up. Everything is a link back to where it already lives, never a copy.

The honest part: today the brand, affiliate and partnership records can be read by anyone who opens a page's source, and the notes we call private are kept off public pages only because no page chooses to show them. So the relationship notes need their own closed home first, readable only by people on a team list who sign in as themselves. That is decision 1. Nothing has been built and no record was changed to write this.

The flow · where each line comes from and where it never goes
WE TALKED ABOUT THEM ONE BRAND CARD · RELATIONSHIP TAB WE TALKED TO THEM Stories and gift liststhe 12 edits; picks already tied to the brand Rec League76 picks carry the flag Gift guide flags13 open, on 8 brands Sale posts (the Sales tab)once posted or shared Posts, guides, newslettergap: few posts linked; no newsletter log Relationship · one brand (example lines) Status: No contact · Reached out · In conversation · Partner · Paused ABOUT THEM 12 Oct · gift list · link 3 Oct · Rec League · link 22 Sep · story · link fills itself, newest first TO THEM 20 Oct · email · Geren · replied 14 Oct · sample asked · Geren 2 Oct · affiliate: earns who, when, a link, next date Next: follow up 27 Oct · a person here links to their People thread Brand emails we getsale scan: 865 threads, 96 brands Emails we sendforward or copy one address Calls, samples, eventsone tap or a screenshot Applications, affiliate statusHome Base list; earns / tracked Partnerships, meetings11 prospects; meeting notes Desk-wide view · all brands, on the Brand Desk Waiting on a reply · Follow-up due · By status · By who A filter on the list you already use, not a new tool Home Base, each morning "Follow-ups due", next to the people ones from Outreach flow The closed home Readable only by people on the team list, signed in as themselves. Built first (decision 1). No path out: canava.co pages, embeds, widgets, feeds, the brand library, client portals, anything a brand or reader can open. The closed home refuses them, so nothing can leak by a page showing the wrong field. Checked after every change, the same way private-client picks are checked. already recorded; fills itself a person on the team adds it another tool's home; linked, not copied a gap today The rule underneath: each line points at where the thing already lives. The tab only holds what exists nowhere else: who did it, when, what happened, the next date.
Build order and size · each step waits for your yes
1 · Lock first · smallClosed home and a team list with just you on it
→
2 · About them · mediumThe timeline that fills itself, links only
→
3 · To them · mediumSale emails, affiliate status, one-tap touches, who did it
→
4 · Status and desk view · smallStatus line, follow-up dates, the waiting list, Home Base line
→
5 · Before anyone joins · mediumOwn sign-in each, roles, who sees what, how access ends
Your decisions · our suggestion filled in
  1. Lock first
    Relationship notes are the most private thing we'd hold about a brand. The records they'd naturally sit beside can be read by anyone who opens a page's source today.
    Lock first: closed home and team list before any note is typed in
  2. The status words
    Your five, lined up with the reply taps already planned on Outreach flow (replied, meeting, yes, not now, closed) so people and brands speak the same way.
    Status: No contact · Reached out · In conversation · Partner · Paused
  3. Who sees what when others join
    Today the tools only recognise you by your email; the curator key is shared, so it can't say who did something or shut one person out.
    Access: own sign-in each; you see all, team adds touches; no shared key
  4. Follow-up timing
    Same default as the people plan, so one morning list covers both.
    Follow-up: one week by default, on Home Base
  5. How emails get in
    Brand emails we receive are already scanned for sales. For emails we send, there's no brand or press label in your inbox to read from.
    Emails: the sale scan plus one forwarding address; never the wider inbox
Added 25 Oct · brand emails that come straight to us

What happened. Annmarie Skin Care wrote three times (11, 14 and 21 Sep) to the affiliate address, offering a box and their own affiliate programme. The 12 Sep morning briefing showed the first email once. Nothing kept it on a list, and the two follow-ups never showed. The sale-email check only picks up sale emails, and the affiliate address is the one address in your inbox with no account tag, so nothing marked these as brand mail. As more brands come to us directly, not only through ShopMy, this will keep happening.

How it would land, going forward (written, not built):

  • Covered addresses: the affiliate address gets its own account tag like the others. A once-a-day check reads only brand emails from a real person sent to the affiliate address, geren@, admin@, hello@ and brands@ at CANAVA, and admin@ at MICOLI. Sale emails, newsletters and platform digests stay out. Never the wider inbox.
  • A private "Brand outreach waiting" list: one line per email: brand, date, which address, and a link back to the email. It lives in the closed home from decision 1, never on the open brand card. A line clears when you reply, say no, or snooze it.
  • A row on Home Base To do: "Brand outreach waiting: N", oldest first, one tap to the list.
  • The brand card only gets public facts, such as "has its own affiliate programme" with the programme link.
  • Nothing is answered or sent for you. This widens decision 5 (how emails get in) from the sale scan plus one forwarding address to these named addresses.

Waiting today: Annmarie's three, plus 16 other brands: 3 direct asks through impact and 13 ShopMy gifting offers that were never opened.

Inbound: tag the affiliate address; daily check into a private waiting list; count on Home Base To do
The detail: what exists, the security reality, the risks, and what each step involves

What already exists and gets reused (read from the live records this morning, 25 Oct):

  • The Brand Desk card already has eight tabs (Your calls, Products, Proof, People and ownership, Where it goes, Sales, Fashion Week, History). Relationship becomes one more. The page is being worked on by another session today, so nothing here touches it.
  • Picks tied to brands: 16,506 picks link to 542 brands; 42 are live. The 12 edits (stories, gift lists, brand edits) and the "Where it goes" tab already know which brand sits where. This is the backbone of "about them".
  • Rec League: 76 picks carry the flag. Gift guide flags: 13 open, on 8 brands. Sales tab: 1,122 sales captured; none posted or shared to Rec League yet, so that line starts empty.
  • Posts and newsletter: the editorial list has 486 pieces but only 5 name a brand, by typed name rather than a link to the card. The newsletter isn't recorded anywhere. Both are gaps; the plan links them as they're published rather than back-filling.
  • Brand emails: the sale scan has 865 email threads from 96 brands, each with a link to the thread. These become "they emailed us" lines automatically.
  • Affiliate: 7 brands earn, 2 are tracked, 63 marked none; 100 entries in the affiliate list (ShopMy and others); 45 items on the Home Base affiliate admin list, many of them applications ("Apply to Naboso…"). An application or a status change becomes a "to them" line.
  • Partnerships: 11 records, all at "prospect", with fit, value and heat; 0 follow-up items. The Fashion Week houses have an ask-and-reply log with who asked and when, still empty. Its shape (asked on, by, what, replied on, what they said) is the shape we reuse.
  • Channels: the four existing words (In person, Phone, Email, Social media) are reused as-is.
  • Outreach flow (23 Oct): the people side of the same idea: one dated line per touch on a person's thread, follow-ups due on Home Base, one-tap outcomes. A brand line with a named person links to that person's thread; nothing is written twice.
  • Older outreach: the Outreach and Strategic Outreach tools keep 24 rows in the LIFE records, also open. Folded in later, then those tools retire. Evolve, don't add.
  • Gmail: labels are by account (CANAVA admin, Micoli) and topic; there is no brand, press or partner label, and SALES holds 1 thread. Nothing to reuse beyond the sale scan.
  • Your decisions log (9 entries) is the one record already closed to everyone but us. It is the model for the closed home.

The security reality, plainly. Private-client picks and the client list are properly locked: a page can't see them unless it carries the curator key, and we check that after every change. Photo, PDF and tools uploads need your sign-in since 21 Oct. But the brand records, the affiliate list, partnerships and the picks can be read, and most can be changed, by anyone who opens a page's source. The "private" note on a pick (30 of them), a brand's notes and its affiliate note are kept off public pages, embeds and feeds only because no public page asks for them. That's habit, not a lock. So, today: yes, relationship data can be kept off public pages, but only if it lives in a new closed home, not in those records.

What's needed before other people get access: (1) their own sign-in, with the email code the tools already use; (2) a team list saying who is on it and in what role, because today the records recognise only you by email, and a few let any signed-in account read everything; (3) who sees what: the closed home checks "on the team list", never just "signed in"; (4) how access ends: remove them from the list and it ends at once; (5) the shared curator key stays yours and is never handed to a team member, because it can't tell people apart. There are 2 sign-in accounts today, both yours. Whether a stranger could create an account on their own I couldn't confirm from here; the plan doesn't rely on it either way.

The risks:

  • Access before others join (the big one). If someone is given the tools before step 5, they could read every open record, change brands, and, if handed the shared key, see client picks. Suggestion: nobody new gets in until step 5 is done and checked.
  • A note typed in the wrong place. Before step 1, the natural place to jot "spoke to their PR" is a brand note, which is readable from page source. Suggestion: no relationship notes anywhere until the tab exists.
  • Copies drifting. Copying a post or an email into the tab means two versions that disagree. Every line is a link; only who, when, what happened and the next date live here.
  • Inbox overreach. Reading the whole inbox for brand names would sweep in personal mail. The sale scan plus one forwarding address keeps it narrow.
  • Name matching. Posts and older outreach name brands by typed name, so some won't match a card. Unmatched ones go to a short To check list, never guessed.
  • The Brand Desk is busy. Another session is in it today. The tab is built only after that work lands, as one more tab, not a new page.

What each step involves. 1 · Lock first (small, about one session): the closed home, the team list with you on it, and the standing check that a public page gets nothing back. 2 · About them (medium, one to two sessions): the timeline reads picks, edits, Rec League, gift guide flags and sale posts, links only, newest first. 3 · To them (medium, one to two): sale emails and affiliate changes arrive by themselves; a one-tap add for email, call, sample, event or application, stamped with who. 4 · Status and desk view (small, one): the status line, the next date, the Waiting on a reply and Follow-up due filters on the Brand Desk list, and the Home Base morning line. 5 · Before anyone joins (medium, one to two): each person's own sign-in, roles, and moving the open brand, affiliate and partnership records to "anyone can read what's published, only the team can change" so a teammate's access doesn't open anything else.

Where this came from. The live records (read only), the Brand Desk and Home Base pages, Outreach flow, the security notes from 20 Jul, 7 Oct, 8 Oct and 21 Oct in CA_BUILD › Active, and your Gmail label list. The four memory notes named in the ask (security state, security lane after 16 Sep, default-closed rule, private client access) aren't in the folders this session can reach, so their content was taken from the security notes and checked against the live records instead. Nothing was built, and no record or setting was changed.

Four areas, six surfaces each: what is settled, what is moving, and where a brief is still missing.

Updated 25 Oct · since the plan: the Brand Desk card now has seven tabs: a read-only Sales tab (from GSOS) sits between Where it goes and History. It also gained a Note to us box, name and website edits in place, and a work-through list that moves card to card. Behind the briefs, the four research jobs (brands, places, resource links, community) now save through new save functions. Each save fills only empty fields, keeps what is already there, and ignores your verdict, rating and voice fields even if sent. Pages are read on the server; finding things is back on Claude's web search. Proven 25 Oct. Nothing a reader sees changed, and both sets of five decisions (the consumer card brief and the Directory) are still open.

Updated 24 Oct · new plan: the Directory, with Brand Desk showing brands only and one desk with a tab for every other role. Written, not built; five decisions for you.

Updated 23 Oct · next step for Brands: the consumer brand card brief is written and waiting on your five decisions. Your backend card (Option A) went live on the Brand Desk today. Nothing for readers has been built from the brief yet.

Each square shows the live link and the brief behind it, if there is one. The outlined square in each row is the one lagging most. Read-only: nothing was changed to make this page. Anything I could not confirm is marked unconfirmed.

Settled · approved design, built and live In progress · built or part-built, design not signed off Brief exists, not built No brief
LibraryGroup viewSearchIndividual cardWidgetsSocial
BrandsFood, style, home, home keeping, pets, movement Most lagging
In progress

The library page is built, but its sign-off has been open since 3 Aug. It shows no brands today: none are published (checked this morning). All 12 were taken down on 27 Sep for full vetting. Two older library pages are still up.

LiveBrand Library BriefBrand library design brief, 24 Jul (a draft, not locked)
No brief

Readers can only filter the library by vertical. There is no group or edit page yet. Brands are grouped into stories, gift lists and beta picks, but only on the Brand Desk.

MockBrand edits, 2 Oct (design only)
Brief exists, not built

There's a search box on the library page, by what a brand makes. It isn't shared with the other areas. The Brand Desk search (tags, keywords, products, 20–21 Oct) is yours only.

LiveLibrary search box BriefSearch & filtering brief, 19 Aug (approval not recorded)
In progress

Your Brand Desk card is settled (card family, approved 13 Oct), but readers never see it. It moved to the Option A layout on 23 Oct (brand card plan). The reader's brand page opens as a side panel with no approved design, and its photos are 4:3 and 16:11, not 3:4. There are 10 different brand cards in use. The consumer card brief was written 23 Oct, below; nothing has been built from it.

LiveBrand page (reader) BriefConsumer brand card brief, 23 Oct (on this page) EarlierBrand library brief, 24 Jul · Brand page brief, 10 Aug · Brand tile finishing pass, 20 Aug · Brand card versions, 13 Oct

Efficiency note on the Desk card, 23 Oct (for the brief; nothing redesigned):

  • Was duplicated, now one: two research controls ("Research this brand" in the top bar, "Ask for research / Cancel the ask" under Where it is). Where it is (added, research, live, Put live) now sits in a strip under the top bar.
  • Still duplicated: verdict shows three times (hero, Your calls, tiles); the product link shows twice per tile (chips and the paste field); Investigate and Gift list each appear as a chip and in the To check list.
  • Buried: Commons, Tags and the Products filters sit below long sections; the nav jumps help, but on phone the card is a long scroll before Products.
  • Little used: Copy embed (beta only); the Story builder link; the "Words waiting" review; old "Where it goes next" mixes.
  • To decide: one action bar (research, live, verdict) at the top; fold empty sections to one line each; Products and Commons as tabs, not sections.
In progress

There are two brand widgets in two different designs. The older one is built on the Art Out East feature card. The newer one is a temporary beta embed, "design pending". Only YUZUCO has a cleared photo. Neither appears on canava.co.

LiveBrand widget LiveTemporary brand embed BriefNone of their own
No brief

A mock from 29 Jul: story units, a four-slide carousel and share cards, with temp photos and sample copy. It isn't connected to the brand record and hasn't been approved.

MockBrand Social Kit
Media DietGuide the audience, build community
In progress

Live, but its design is a placeholder (approved as a placeholder 1 Oct). You turned down two August directions. Approval of the Art Guide reframe isn't recorded. 37 titles are live, and 1 has a cleared image (checked this morning).

LiveMedia Diet reader MockMedia Diet system, 1 Oct BriefDesign direction & library direction, 8 Aug (both turned down) · Art Guide reframe, 8 Aug
In progress

Mixes work: you can make them on the Desk, and they show as tabs on the reader. There's no designed mix page of its own yet.

LiveMixes on the reader BriefMedia Diet kit spec, 21 Jul (proposed)
Most lagging
No brief

Readers have no search anywhere in Media Diet; only the Desk has one. A guide people can't search can't connect like-minded people, and that's its purpose. The only cover is the cross-area search work from 19 Aug, which was an assessment.

BriefNone for Media Diet
In progress

The Desk card is settled (card family, 13 Oct). The reader card has two competing patterns: a side drawer on the live reader, and a tabbed pop-up on the beta. The 3:4 photos from the 13 Oct pass landed. The rest of that pass is still a draft.

LiveDrawer (live reader) BetaTabbed pop-up BriefMedia card design pass, 13 Oct (draft)
In progress

There are two embed systems, and neither has a design. The widget uses 4:5 and 4:3 frames, not 3:4. It reads the full title list rather than only the cleared ones, so check that it can't show an image that isn't cleared (unconfirmed).

LiveMedia widget LiveTemporary media embed BriefNone
No brief

The only social piece is a placeholder Instagram portrait card inside the widget. There's no carousel and no share card.

BriefNone
Travel / JourneyThe furthest along; the model for the rest Most lagging
No brief

Readers have no page that gathers the guides. There's no brief, no mock and no page. A travel overview map and a stays guide reader are both noted as not built.

BriefNone
In progress

The map engine is settled. The guide as a page for readers has only been designed once, for Art Out East. Cover image and intro on every guide, and a front page for Out East, are still to do.

LiveArt Out East on canava.co LiveReader map BriefMap redesign brief, 7 Jul · Art Out East design handoff, 18 Jul
Settled inside a guide

Search, Near me and the compass all work inside a guide; the 19 Aug brief rates it the best search in the system. Search across guides isn't built: canava.co's site search can't see the guides.

LiveSearch in the guide map BriefSearch & filtering brief, 19 Aug (covers search across guides)
Settled

The map's place card has its 3:4 photo with swipe (9 Oct), directions by name, and the PRO TIP. Your curator card (the 9 Sep version) is settled too. The stay card is built from the feature card, with square corners, so it doesn't match the place card.

LivePlace card (open any pin) LiveGuide Curator card BriefClaude Design map embed handoff, 7 Jul
In progress · built, not refined

The Art Out East widgets work, and so does the place widget, which has five layouts. The place widget's agreed trim isn't in the live version. The 8 Oct update was never confirmed live.

LiveWidget catalog LivePlace widget BriefPlace widget scoped pass, 27 Jul. No refinement brief.
Brief exists, not built

Sharing a place works: link previews and the Share button. Guide and trip share cards, and Share as image for Stories, are briefed but not built. You're working on this now.

BriefShare & distribution brief and playbook, 8 Jul
Food / RecipesCard and flow far along; listing behind
No brief

There's no CANAVA recipe index, only the standard Squarespace food listing. The recipe card links its course and diet labels to an index page that doesn't exist, so those links go nowhere.

LiveSquarespace food listing BriefNone (the 19 Aug implementation map lists it as missing)
Brief exists, not built

The path (Food › Course › Main group › Anchor ingredient) was approved on 14 Oct and works on the Desk. Readers have no page that groups recipes. Most of the 37 suggested paths are still suggestions.

LiveRecipe Desk (yours) BriefRecipe taxonomy & search proposal, 14 Oct (approved)
Brief exists, not built

Readers use canava.co's site search, through a third-party plugin. There's no recipe filter for readers. The search words behind the Desk are built.

Livecanava.co search BriefSearch & filtering brief, 19 Aug
Settled

The approved recipe card (19 Aug) is the full reader embedded on canava.co. You checked all 18 live pages. An older copy and a test version are still up.

LiveRecipe card BriefApproved recipe card, 19 Aug · Recipe design brief and embed spec, 19 Aug · Recipe flow design pass, 12 Oct
In progress · built, never reviewed

The card, teaser and tile layouts work. The look hasn't changed since about 13 Oct, and its design has never been reviewed. Since 15 Oct the Desk offers it only as "Small card instead".

LiveRecipe widget BriefNone of its own (the lock sheet's step 12 names it)
Most lagging
No brief

There's no brief, mock or build. The Desk has one line saying a social post is "one more view of this same record". The smoothie posts from 16 Sep are drafts, not a design.

BriefNone
The level set

Where each area stands.

Brands

Strong: the Brand Desk card (card family, 13 Oct) and the entity record behind it.

Lagging: everything a reader would see. The library has no approval and no published brands. There are 10 brand cards, two widget designs, no group page, and social exists only as a mock.

Most lagging: the Library. It's the backbone, and today it's empty and unsigned.

Next step (23 Oct): your five decisions on the consumer brand card brief. The card, tile and embed all come from it, and the library uses the same tile.

Media Diet

Strong: the Desk card, mixes that work, and a live reader.

Lagging: the reader's design is a placeholder. The record has two ways to open (drawer vs pop-up), the two embeds have no design, and readers have no search or social.

Most lagging: Search, the surface its community purpose depends on.

Travel / Journey

Strong: the map, the place card, search inside a guide, sharing a place, the Art Out East widgets and the area chain.

Lagging: the guide as a finished page (designed once, for Art Out East), widget refinement, and guide and trip social.

Most lagging: the Library, a front door that gathers the guides.

Food / Recipes

Strong: the approved recipe card, live on 18 pages, and the approved course path on the Desk.

Lagging: nothing that lists or groups recipes for readers, the widget's design has never been reviewed, and there's no social.

Most lagging: Social, with nothing at all. A close second is the Library, because the card's labels already point at an index page that doesn't exist.

Gaps between areas

The most lagging surface overall: the Brands Library. Brands is the backbone of everything outside editorial. Its library has been waiting on sign-off since August and shows nothing today. The brand card, group view, widgets and social all depend on it.

Travel as the model

What Travel can lend.

Travel is the furthest along, and it's what the other verticals will draw on as they fold out to publish. Here is each pattern: whether it's ready to reuse, where it lives, and what it could serve.

PatternReady to reuse?Where it livesCould serve
The area chainGuides nested inside guides: town, county, region. Settled. Locked 1 Jul, and the Upstate zones were locked 9 Aug. It already works for any vertical, with no destination built in. The guide dashboard breadcrumb uses it. The new town › county › state chain on curator cards (22 Oct) comes from a different source. Guide Dashboard Brand groups (Food › Pantry › Oils), Media Diet mixes inside mixes, and recipe collections by region or season. The shell note already names a shopping guide and a media guide as future users.
Art Out East widgetsFeature card, teaser tile, text tile, title band. Settled. Shipped 25 Jul, and not tied to any guide. Your own images and copy go in through the Widget Editor. They need a refinement pass, not a redesign. Widget catalog · Feature card · Teaser tile · Widget Editor The base for a single widget family. The brand widget already uses the feature card. The recipe and media widgets are separate designs that could fold in.
Place cardsThe map card, the curator card, the place widget. Curator card settled (the 9 Sep version is the source of the card family). Map card and place widget still being adjusted (9 Oct, 8 Oct). Guide Curator · Map card · Place widget A single reader card for brand and media, with the 3:4 photo, the facts and the PRO TIP. The card PNG export is already set aside for the brand library.
The guide mapThe map with its search band, Near me, place filter row and named lists. The engine is settled, but it's still changing daily (8 reader updates on 20 Oct). It's mostly generic, but a few guide names are still built in, including Art Out East as the default. Reader map · Guide Map (yours) · Art Out East on canava.co Brand stockists and where to buy, and recipe journeys (farms, markets, makers). Not Media Diet: you turned down a map-shaped Media Diet on 8 Aug.
Journey mapsThe residency board and the research ladder. Not ready as is. The board is wired to the Maine trip alone. Its card-opening bug was reported three times. The research ladder (Journey Drafts) works for any guide, but it has never been used to move a guide to ready. Journey Board (read-only) · Journey Drafts The idea carries over: a board of a plan by place and date. For recipes that could mean a season or a trip's cooking. For brands, a buying trip. The board itself would need to be made general first.
Client portal tilesThe Compass: a private page made of modules for one client. Still moving (12 updates on 20 Oct, the journey layout on 20–21 Oct). It already mixes travel, recipes (the recipe hero, 13 Oct) and products. It's private: each client opens it with their own link. Private Clients (each portal opens from there) A member or client edit across all four areas: a trip, its recipes, its brands, what to read. The private-client framework (19 Jul) and member access plan (20 Jul) are still drafts.
What Travel still needs designed
Brands · the reader card brief

The consumer brand card.

A design brief, 23 Oct 2026. Design only: nothing here has been built, and no tool or record was changed to write it. The Brand Desk card you rebuilt is your backend; this is the card a reader sees.

What already exists, in three lines

Done: a lot of design work for readers. The library brief (24 Jul) set the tile and six tabs, and the brand page brief (10 Aug) set the order, the answer coming first and the honest money. There are eleven mocks (29–30 Jul), a finishing-pass prompt for Claude Design (20 Aug), and a working tile-and-panel page that isn't on canava.co.

Not what they look like: the two 20 Sep papers aren't about the reader card's design. They cover whether the 136 fashion week houses join brands (A merge / B link / C keep apart) and the flags on your own card, and your ruling isn't recorded.

Not done: no reader card design has been approved, there was no brief written since the 3:4 card family, and no brand card or embed is on canava.co. The reader card note further up this page was a recommendation until this brief.

Who it's for

The reader first. Curious and scanning, usually on a phone. They arrive from a story, a guide, a gift list, an embed or a shared link, and want to know fast whether to trust this brand and where to get it.

The friend you hand one brand to. The card has to stand on its own, with no library around it.

Later, members and private clients, who save and shortlist. Leave room for Save, as the map card does with Save to trip, but don't design the account side yet.

Not you. Your working, flags and private notes stay on the Desk card.

The rules it's held to
  • The lead photo, with the brand beside it. Never a small photo, then empty space, then big type.
  • Colours from Brand Standards: cream, navy, and the four verdict fills, which are the only strong colour.
  • IvyPresto light for the name only. Helvetica for everything readable.
  • Answer first; depth below. Sections with nothing to say hide.
  • No overlay, blur or tint on photos. Logos shown whole.
  • Plain words, never field names. Checked against Design Rules before you see it.
What the reader sees, in order (phone first)
  1. The lead photo and the brand together. A 3:4 photo. On desktop the record sits beside it, as on the card family and the map card. On a phone the photo runs full width, with its height capped so the name, verdict and one line land on the first screen. With no cleared photo, the name set large as a deliberate type card, never a grey box.
  2. Beside the photo: the name, the one-line read, the verdict said once (sage, dusty blue, slate or coral, with its gloss: "Go. No hesitation." and so on), what it makes, and who owns it: founder-led, family, PE- or VC-backed. Ownership sits up here because it's the CANAVA signal.
  3. Our take. Your published words: why we land here, in a few lines, with no box around it.
  4. Known for · New. The products you marked, as small 3:4 tiles with a name, a price and a link. Known for first. Up to four of each, then "See all".
  5. What others say. Outside ratings and certifications, each with who said it and when it was checked. If none have been checked, one quiet line: "Not checked yet".
  6. Made and sourced. Where it's made, materials, and the seven values with any unrated ones dimmed as "To assess" (see decision 4). This hides entirely when nothing applies.
  7. Who's behind it. Founders and owners, as names you can tap when they have their own record.
  8. Where you'll find it on CANAVA. The stories, edits, guides and gift lists it's in, then "What we'd point you to instead", each linking to its own card.
  9. Where to buy. The brand's own shop, then stockists (with a map when there are shops to visit), then the links.
  10. The quiet footer. The affiliate disclosure in one line, shown only when a link can earn (your 13 Oct rule), then sources and the date it was last read.
Public on the cardBackend only, on the DeskOn the record, not public today: yours to call
Your verdict (once approved) and your published words. Outside ratings and certifications with source and date. Where it's made, materials, values. Products marked Known for and New, with price. Where to buy, with links and the disclosure. Founders and owners. The stories, edits, guides and gift lists it's in. Alternatives. Sources and the date read. Your private notes. Flags and what's owed. A proposed rating until you approve it. The "not looked at" list. Commission rates and affiliate status. Pipeline, keywords, design favourites, review notes. Outreach records, until you rule. Investors and funding rounds, the debate, impact numbers, loved products, where it's sold, social handles. These are on the brand record but not in what the public pages can read (checked this morning). Each one needs a yes before it shows.
One card, three sizes

Full card. All ten parts above. In the library and group pages it opens over the list: a side sheet on desktop, a bottom sheet on a phone, the same mechanism as the map card. It also has its own link, so a story or a share opens it directly and it can later stand as a page of its own.

Tile, for the library, a group or edit, and search results. The 3:4 photo, then name, verdict and one line. Nothing else: no chips and no buttons. The same tile in every grid.

Embed or widget, for canava.co pages, in three cuts from the same card: the tile; the compact card (photo beside the record, which is the temporary beta embed's layout grown up); and the full card as a frame that sizes itself, as the recipe card does. This replaces the two brand widgets and the extra brand cards once you've approved it.

Phone first: design at 390 wide, then widen. Pill rows stay on one line on desktop.

How it reuses Travel and the 3:4 family
  • Card family (13 Oct): the 3:4 photo with the record beside it, the type scale and labels. Readers get the calm version of the layout you already approved for the desks.
  • Media Diet 3:4: the same frame and tile, so a brand and a media title sit side by side in a mix or a story without looking like two products.
  • The map card: photo swipe, the sheet that opens over the list, the PRO TIP box (here, "Good to know"), and Share with link previews.
  • The guide map: Where to buy draws stockists on the reader map, so a brand gets a map without a new map.
  • Art Out East widgets: the feature card and teaser tile cuts for canava.co pages.
  • The area chain: group pages by vertical, nested (Food › Pantry), the same way guides nest.

The shared skeleton for all four areas: photo, then the record beside it, then our take, then sections that hide when empty, then where to find it, then a quiet footer. Brands, media titles and places all fill it. The recipe card stays its own approved object and borrows the tile.

Five decisions for you
  1. The photo on a phone3:4 at full width is about 520px tall on a phone, which pushes the name off the first screen. Suggestion: full width with the height capped so the name, verdict and one line sit on the first screen. On desktop, keep the photo beside the record.
  2. Corners: square or about 5pxThe August brand brief and the Art Out East page are square to match canava.co. Design Rules (3 Oct) and the card family use about 5px. Suggestion: 5px, so one system runs across all four areas, and settle the hub at the same time.
  3. Which of the held items go publicInvestors, the debate, impact numbers and where it's sold are on the record but not public. Suggestion: show investors and funding when sourced, because they feed the ownership signal; keep the debate for stories; show impact numbers only when each has a source.
  4. The seven values when most are unratedAs of 20 Sep only 32 of 2,612 brands carried any source, so most cards would show "To assess" on nearly every value. The August mock says "quiet presence, never a blank". Suggestion: show the values block once at least one value is rated, with the rest dimmed; hide it when none are.
  5. The fashion houses (A, B or C from 20 Sep)This decides whether 126 researched houses get a brand card, and whether per-fact sources ("what they say, what they don't") appear on it. Suggestion: B, link, as the paper recommends. The reader sees only approved facts in the house's own words, and the outreach line stays off the public card until you say.
Updated 25 Oct · Brands and everything else · the Directory plan

One home, many desks.

A design plan, 24 Oct 2026. Written, not built. Nothing was moved, re-labelled or deleted to write it. Counts were read from the live records this morning.

Updated 25 Oct · what changed

You said yes to moving the clear ones, and they're moved. The test was re-run on today's 5,359 cards, so the new ones were included: 680 came out clear. 675 moved; 5 were held back for you.

Where they went. 593 to Resources: 81 certifiers, 207 investors, 260 owners and groups, 39 nonprofits and 6 platforms. 72 to places on their guides: 62 places and 10 museums and venues filed as nonprofits, each already on its guide, so nothing new was added there (two, Aperture Foundation and Ursa Major, still have no pin on the map, as before). 10 to the Media Desk: 5 were already titles there and are now linked; 5 show in a "From the brand cards" line at the top of the Desk. None went to People: no person came out clear.

Kept both. 24 are also real product brands, so they keep their brand side and stay on the Brand Desk too: for example Rachel Comey, Ursa Major, Canyon Coffee, Sey Coffee, Seed Savers Exchange, Camper, Kvadrat, Momofuku, St. Agni and Whole Foods Market (its own label). No card lost the brand role. 58 gained a role word their own links prove (certifier, investor or owner).

Held back for you (5): Kitava, Roon and Suno carry the brand role and may or may not sell their own products; The Salad Project has no place on a guide yet; L'Impatience's place isn't on a guide yet. They're at the top of the review list.

How a move works. The card stays one record and keeps its links, products, notes and your calls. It still opens from its old Brand Desk link. Each desk shows it by where it lives now, and links back to the one card: Resources has a "From the brand cards" tab, the Media Desk and People have a "From the brand cards" line, and places are the guide places they were already linked to. Nothing was copied or deleted, and the cards' last-changed dates weren't touched.

Review in batches. The likely (550) and to-check (365) cards, the 5 held back, and anything you flag with the new "Not a brand" tap on the Brand Desk are in one list: Not a brand? · review in the Trawl Log, 25 at a time, with one tap for place, person, resource, media or keep as brand, and an Undo on each. Home Base To do has a "Not a brand? · review" row with the live count, the oldest date and the link. New cards that look like non-brands join the list each night; they're never moved without you.

Undo. Every move is logged with before and after. Any one can be put back from the review list, and the whole 25 Oct batch can be put back in one step if you ask.

Still to do. The Brand Desk list still shows the moved cards until its own session hides them (it's busy with the "Not a brand" tap tonight, so I left it alone). The older "Not a brand?" view on the Brand Desk still counts them too; the review list above is the working one.

In 30 seconds

The one table stays. You were right on 22 Sep: one home means every future tool can draw on it. What changes is how it's shown. The Brand Desk shows brands only, with products, research and catalogues untouched. One new desk shows everything else, with a tab for each role. A card with two roles is one record that appears in both tabs. The Brands tab opens the Brand Desk card.

What the records say today: there are 4,988 cards. 941 carry the brand role, and 2,767 carry no role at all; that's the biggest surprise (decision 2). 238 carry more than one role. The "Not a brand?" list flags 1,567: 659 clear, 550 likely and 358 to check. None of the 4,988 is published.

The flow · how a new name comes in and where it lands
COMES IN ONE CARD, ITS ROLES LANDS IN You add it+ New brand, Capture, a screenshot Trawla capture filed as a company Intake sessionsyour lists, ShopMy, sweeps Research names itan owner, investor, stockist, maker Sale scan, certifier readingnames found in emails and ratings One card, one home Found by name first, never made twice. A name in a sentence becomes an empty card. Empty is fine. Its roles, from your role questions "Does it put its own name on what it sells?" brand "Does it sell other companies' things?" retailer "Has it put money into another company?" investor Research proposes; you confirm. Several is normal. No role yet goes to To sort, never guessed. To sort Clear: fixed in bulk only with your OK. Check: a filtered list for you. Nothing deleted. Brand Desk · brands onlyproducts, research, catalogues, product scan, Put livethe only place a card can go live for readers The Directory · one desk, a tab per role All · Brands (opens the Brand Desk card) · Houses Owners and groups · Investors and funders Retailers and stores · Platforms · Nonprofits Certifiers · Makers and manufacturers · Media Places to route · To sort Each tab: its own card and its own research set. Two roles = one record in two tabs. Nothing here is published. Research for these tabs queues after your brand asks, in the same 1am run. Guide Curators, by locationa place goes to the guide forits town, with your OK Media Desk · Peoplepublications and persons keeptheir homes; the card links Resources, Trawl, Design referencekeep their notes; the Directory card links to them, never copies yours a session or the nightly job; you check another tool's home waiting on you The rule underneath: the card is the one home for a company's facts. Desks and tabs are views of it. Moving a card means changing a role, never copying a row. Places follow the standing rule: a place lands on a guide only with a guide entry, its pin and the guide's own place word, not as a bare row.
The tabs · what each shows, its research, and whether it's ever public
Tab · who's in it (today)The card showsResearch setPublic?
AllEvery card · 4,988Name, its roles, which tabs it sits in, how many links it has, and any To check items.None of its own.Never.
Brandsbrand role · 941 (plus decision 2)Opens the Brand Desk card as it is today: Option A, products, Proof, People and ownership, Where it goes, Sales (read-only, from 24 Oct), History.The full load-up and the product scan, unchanged.Only through Put live on the Brand Desk (0 live today).
Houseshouse · 345The fashion week grid: the eleven questions, each fact with its source, date and state; the season appearances; on or off schedule.The eleven-question disclosure set.Never from here. If a house is also a brand, it goes live through the Brand Desk.
Owners and groupsowner 218 · group 36What it owns, what it's part of, its ownership type, people, and the brands under it with their verdicts.Map the money, ownership half: parent, holdings, sources.Never. Its name shows on brand cards as "who owns it".
Investors and fundersinvestor 170 · funder 48Investor type in your words (venture capital, corporate venture, private equity, angel, family office, hedge fund, institutional, founder's firm), its portfolio (780 investment links today), funds and their size, focus, partners, your verdict on the firm, and its Lab brief.Map the money (from 3 Oct, Forerunner first). Funders: who they give to and on what terms.Never.
Retailers and storesretailer 321 · store 133 · distributor 12What it stocks (linked brands), whether it has its own label, its locations (a store has a pin), how it sells, and which network pays.Where it sells and what it stocks. A catalogue only for its own label.Never from here. A store reaches readers as a place on a guide.
Platformsnew role · 46 flaggedThe eleven sections from the Vette brief: model, consumer flow, economics, legal and money seats, data, paid placement, vetting, technology, people, comparisons, corrections. An empty section is a To check, never blank.The platform set (Vette, Spring as the comparison).Never.
Nonprofitsnew role · 121 flaggedMission in its own words, its programmes, its funding (who funds it and whom it funds), people, and the brands it partners with or benefits.Mission, programmes, funding, each with a source.Never.
Certifierscertifier 59 · membership body 22What it certifies, its reading and weight (one reading feeds every brand that cites it), and the brands citing it.The certifier reading queue that already exists.Never as a card. Its reading shows on brand cards.
Makers and manufacturersmanufacturer 81 · mill 4 · grower 3 · factory 1Who it makes for (named-supplier links), where, what materials, and how open it is (established, opaque).Supply-chain facts with sources.Never.
Medianew role · 129 flaggedWho owns it, plus a link to its Media Desk title when it has one.None here. The Media Desk researches titles.Never from here. The Media Desk publishes.
Places to route164 places and venues · 6 eventsName, town, and the guide it would go to (the list names one for each, across 34 guides), with a "Send to the guide" button that waits for your OK.None here. The guide side geocodes and checks it.Only once it's on a guide and published there.
To sortno role, or flagged "check"The reason it was flagged, its products if it has any, and one-tap role choices.None until it has a role.Never.
Cleanup of the 1,567 · nothing deleted
  • Clear (659): certifiers 81, investors 207, owner groups 239, places 65, nonprofits 49, media 12, platforms 6. You see the batch by kind and say yes. Each card gets its right role added, and loses the brand role only where it had one. Every change is logged with before and after, and can be undone.
  • Likely (550): the same, a kind at a time, with each row shown so you can untick any before saying yes. This includes the 222 shops marked "her call" (with 156 products on them), which you look at row by row.
  • Check (358): a filtered list in To sort, never bulk. 200 are unclear, 41 shops, 35 owner groups and 32 people. 496 products sit on these cards, so none of them leaves the Brand Desk until you say.
  • People (32): they go to People (research) only with your OK. The card stays, linked.
  • Places (164): "Send to the guide" adds a guide entry with its pin and the guide's word. The card stays and is linked.
Brand Desk · what changes, what stays

Changes: the list, the counts and the search show cards with the brand role, plus the no-role cards (until decision 2 is settled). A card with other roles gets one line, "Also in the Directory: investor, owner", with a link. The "Not a brand?" button becomes "To sort" in the Directory. A new brand gets the brand role when it's made.

Stays the same: the Option A card and its tabs (seven since the Sales tab, 24 Oct). Products, Known for and New, the product scan, catalogues, Pull link. Research this brand and the 1am run for your asks. Verdicts and Put live. Every link to a card, so the team's bookmarks keep working (a non-brand card opened by its link still opens, with its Directory tab named).

For the team: nothing they do today moves or disappears. The only thing they'll notice is a smaller list.

Where other tools already hold organisations · link, don't copy
Build order · rough size
  1. Your five decisions below.
  2. Add the missing roles (platform, nonprofit, media), each with its own question, like the others. Hours.
  3. The Directory as a shell: All and To sort, read-only. About a day.
  4. Cleanup pass, the clear ones first, with your OK and undo. Half a day, plus your review.
  5. Brand Desk shows brands only, with the "Also in the Directory" line. Half a day.
  6. The tabs one at a time: Investors and funders, then Owners and groups, Platforms, Certifiers, Nonprofits, Retailers and stores, Makers, Media, Places to route, then Houses once the fashion house question is ruled. About a day each.
  7. Research sets per tab, added to the 1am run after your brand asks. About a day, then nights.
  8. Links to Resources, the Media Desk, Trawl and People. One to two days.

Two to three weeks of sessions in all. That's an estimate, not a measure.

Risks
  • Brands vanishing from the Desk. 2,767 cards have no role, 149 of them with products (8,301 products in all). A "brand role only" filter would hide them, which is why decision 2 comes first.
  • Research writing brand fields onto a non-brand. The nightly queue now treats a no-role card as a brand. Each tab needs its own research set before its cards are researched.
  • Something going live that shouldn't. Put live has to stay brand-only.
  • Doubles. Resources, the Media Desk and the cards overlap. Link through the name checks; never merge quietly.
  • The team mid-task. Bookmarked cards and lists. Keep every card link working.
Five decisions for you
  1. The nameThree that are already in use: Directory (the working name; the tool hub is already called the master tool directory), Organisations (from the 13 Oct investor plan and the "organisations home" note on the Not-a-brand list), and Roster (the fashion week roster). Suggestion: Directory, because it also covers places to route and media, which aren't organisations in the plain sense.
  2. The 2,767 cards with no role2,183 of them aren't flagged by the Not-a-brand list. Research already treats a no-role card as a brand or shop. Suggestion: with your OK, give the 2,183 the brand role in one logged, undoable pass. The 584 flagged ones go through the cleanup instead.
  3. Three new rolesPlatform, nonprofit and media aren't roles yet. Adding a role takes one line, with a question like the others ("Does it run a shop for other people's things?", "Is it a nonprofit or foundation?", "Is it a publication or channel?"). Suggestion: add all three. Events go to Places to route, and schools or institutions go to Nonprofits.
  4. Shops that also sell their own labelThere are 263 shops, 298 products among them. Suggestion: a shop with no label of its own leaves the Brand Desk. A shop with its own label keeps the brand role too, so it sits in both, as one card.
  5. HousesThis is the 20 Sep A/B/C question again. Suggestion: B. A house is a brand at a fashion week. It lives in the Houses tab with its evidence grid, and gets the brand role once it has a shop.
Briefs to write next

In order.

  1. The library and group page, one design for all four, starting from the Travel guide pageCovers the brand library and its groups, a recipe index and course pages (which also fixes the broken links), Media Diet mixes, and the front door for the guides. Cover, intro, filter row, grid, and how a record opens.
  2. One reader card and one way to open a recordThe brand half is now written (consumer brand card brief, 23 Oct). What's left is the media card (drawer or pop-up), on the same skeleton, then retiring the extra brand cards once you've approved a design.
  3. Search readers can use, shared across areasOne search box and one set of filter rows, starting with Media Diet, which has none. Decide whether canava.co's site search stays, given it can't see the tools site.
  4. Social, built from Travel's share patternGuide and trip share cards and Share as image first, then recipe posts, then the brand kit (from the mock) and Media Diet portraits. Every piece draws on the same record.
  5. Widget family refinementOne review across the place, Art Out East, stay, brand, media and recipe widgets: one frame ratio, one corner, and the cleared-image rule everywhere.
  6. Making the journey board generalLift it off the Maine trip so recipe and brand journeys can borrow it. Do this once the guide page and social are settled.

Gathered 23 Oct 2026 from the build notes, briefs, tool histories and the tool files, without opening any page in a browser. Brand and media counts were read from the live records this morning. Nothing was edited or moved to make this page; the one change is a tile pointing here from the design home.

Briefs without a link are working documents on your computer, in the CANAVA_COWORK folder, named as shown. Most are under CA_BUILD › Active (some in its Design Brief Package). The July and August brand briefs are in the Reports folder at the top of CANAVA_COWORK.

Unconfirmed: whether canava.co embeds any brand or Media Diet piece, whether guides other than Art Out East and the Kingston area are on canava.co, your brand card ruling from 20 Sep, and whether the 19 Aug search brief was approved.

Status line standard added 23 Oct; see the Workflow Map.