Capture to measure
From forty open tabs
to six good paragraphs.
Seven stages, in one direction. A link enters on Monday from wherever you were reading, and leaves on Saturday as a paragraph somebody read. Nothing in the middle is a scramble.
Curation is a volume activity.
Every stage is built for that.
Not for the demo where three perfect links become a beautiful issue — for the Saturday where sixty items have to become six in ten minutes.
Eight ways in, one inbox
- You never open the app to save something. A bookmarklet, a chat channel, an assistant, a watched feed, a capture key today; a share sheet, an email address and a browser extension to come.
- Everything deduplicates on arrival. Tracking parameters stripped per host, shorteners followed, near-duplicates offered as a suggested merge rather than merged for you.
- Feed entries stay out of the inbox. They arrive as suggestions in their own lane, so the inbox keeps meaning “things I chose.”
Every link arrives legible
- The save never waits on the fetch. Title, author, publication, date, standfirst, hero image, language and reading time are pulled in the background.
- Reader view, so you can judge it. The article's text without the navigation and the subscribe-nags, and without hitting the paywall a second time.
- A page that cannot be read still keeps the link. The item survives with its URL, your note, and a plain explanation of what happened.
A hundred items without touching the mouse
- Keyboard-complete.
jandkto move,1–9to file into a publication,xto drop,uto undo,oto open. - Every destructive action is undoable, and shortcuts never fire while you are typing a note.
- Your note outranks the site's blurb. One sentence on why you kept something is worth more than the article's own marketing copy, so Curio shows it first and never overwrites it.
- One link can go to several newsletters, and read completely differently in each.
Several newsletters, each with its own voice
- A publication is one newsletter: its own sections, schedule, template, voice and mailing list. Pause one without touching capture.
- Sections are buckets you name — “Long reads,” “Tools,” “One more thing” — with optional guidance the writing follows and an optional cap so one section cannot swallow an issue.
- Issues assemble themselves. Everything filed since the last issue, ordered by section, with a draft blurb waiting on each. Assembling twice never produces two issues, and an empty week produces none at all.
A first pass in your voice, never the last word
- Written from why you kept it. Your capture note carries more weight in the prompt than the article does. A model given “the third chart is the whole argument” writes something no summary would.
- Voice profiles. Describe your tone in your own words and give examples of blurbs you consider correct. Versioned, so changing it never rewrites what is already drafted.
- Grounded, or honest about it. Drafts are written only from the piece itself; when there is not enough to say anything true, Curio says so and leaves the blurb to you.
- You can see how much is yours. Every blurb is marked drafted, drafted-then-edited, or written by you — the only honest measure of whether the drafting is earning its keep.
Deliberate, and impossible to send twice
- Pick the list, do not retype it. Choose from the Herald mailing lists you own. A list you do not own cannot appear, because Herald decides that.
- The preview is the email, rendered by the same engine the send uses, so what you approve is exactly what arrives.
- A confirmation that names the list. Not “are you sure?” — the thing worth checking before an unrecallable action is where it is going, so the dialog says it, and the send is refused if it changed while you were reading.
- Send now, or schedule it. Either way it is a decision you make. Curio assembles automatically; it never sends automatically.
Which links did people actually read?
- Opens and clicks per issue, pulled back from Herald automatically — more often just after a send, rarely once the numbers settle.
- Clicks per link, ranked per issue and carried back to the item, so next week's triage knows which of the six things people opened.
- The full archive. Every issue kept as sent: its items, its blurbs, the template version it used, searchable alongside everything you have ever saved.
- “Already sent this?” Capturing something you have published before tells you so, with the issue and the date.
A crash halfway through publishing
cannot produce a duplicate send.
Curio records the created draft before it asks for the send, so a retry resumes rather than repeats. The one failure in this product that could not be undone is the one it is built hardest against.
___PRESERVE_BLOCK_14___
Everything else is undoable.
This one is not, so it is guarded.
Dropping an item, filing it into the wrong publication, overwriting a blurb, merging two links that were not the same story — all of it comes back with u. A send does not, which is exactly why the send is the one thing no agent, schedule or background job in Curio can perform.
Monday to Saturday.
Monday to Friday
You read. Anything worth keeping goes in from wherever you were reading, with one sentence on why.
Friday night
Curio assembles the issue: everything filed since the last one, ordered by section, a draft blurb waiting on each.
Saturday, ten minutes
Reorder, edit the blurbs that need it, write the intro, drop one back to the queue for next week. On a phone, at 360 pixels.
Saturday, one dialog
The preview is the email. The confirmation names the list. You send it, or you schedule it.
Keep the good bits.
Send them Saturday.
Curio does the half of a newsletter that is not sending: capturing, triaging, drafting, assembling, reviewing. Herald owns the subscribers, the opt-in, and the send.
Curio is in active development at Renkara. Nothing here is for sale yet and there is no list to join — write to contact@renkara.com if you want to hear when it opens.
