Archer Scaling AI

Founder Media Engine

Review flow demo, prepared for Chad. Enter the passcode Raine gave you.

Archer Scaling AI · Founder Media Engine
Review demo

One item, the way you would review it on your phone

This is the Your review stage from the call document. Read the script with its angle and sources beside it. Edit any paragraph in place, select a passage to attach a note, leave an overall note, and send the batch once. Then see what the engine does with it. The item below is a sample: the draft, sources and numbers were written for this demo. The workflow is the real one.

Script · LinkedIn post · version 1

Listeners are counted. Relationships are not.

Pillar Artist data Format LinkedIn text post Item 2026-10-05-listeners-vs-relationships QA: 1 warning
Click into any paragraph to edit it in place. Select any words to attach a note to just that passage. Nothing leaves this page until you send the batch.

editedA streaming dashboard can tell you how many people played a song last night. It cannot tell you which of them would drive two hours for the next show.

editedIn today's fast-paced music landscape, it's crucial for artists to leverage data-driven insights to unlock deeper fan engagement.

editedA play is an event. A relationship is a series of events with a name attached: the first listen, the follow, the ticket, the reply to a post, the second ticket. Platforms record the first and throw away the rest.

editedWe are building Liso around the second kind of record. Every interaction between an artist and a listener gets written down as a typed event, with a date, so the relationship can be read back later instead of guessed at from a chart.

editedIf you manage artists and you have ever had to guess which fans would actually buy, I want to hear how you guess today.

Overall note

Feedback on the whole item rather than one passage. It goes back to Script with the batch and becomes a rule candidate.

Your batch

What goes back in one send

Nothing yet. Go back to Review, edit a paragraph or select a passage and add a note.

Approve schedules it. Edits are applied verbatim. Notes on an approved item are filed as rules and do not hold it up. Send back returns it to Script with your notes, and the next version shows you the diff. Reject stores the reason.

Decision record

Approved

Script · version 2 · your edits applied verbatim

Rules extracted

Every correction becomes two things

The change itself, and a learning. The rule is filed into the set the next draft is written against, so a correction is made once instead of every week.

In this demo the rules are extracted by a word diff and a short banned-word list, so you can see the mechanism. In the engine an agent reads the same delta, writes the generalised rule with a sentence of reasoning, and you can edit or delete any rule it files.

Files written

Plain files in the knowledge-base repo, nothing hidden in a database

    
        
    No changes yet. Edit a paragraph or select text to add a note.