Back Office · office.temerarii.xyz
One asset, all the way in — composition, the wireframe + storyboard, the output format stack, and the template, all read from the SAME content-index record. The expected output matches what /media surfaces for this post.
post longform-W46-Sunkind longformweek W46date 2026-11-15campaign longform-youtubepillar strategic_relationsbeat asset videoduration 235.4sground redscenes 10

Checklist the per-video bar — engine/sim

100.0/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W46-Sun good|bad)
✓ all static checks pass — one-focal/scene · tier-by-beat · one-track caption · colorway · cast+shape correct · no banned/fabricated. (audio + visual tiers verify on the rendered finals — Phase 2)

Composition comp · template family · expected output

composition LongFormChaptersfamily / template LongFormChapters
9:16 Reelpending1:1 Squarepending16:9 Widepending9:16 4Kpending1:1 4Kpending16:9 4KpendingGIF (SMS)pending
render pending — silent master not yet on disk
expected output: 0/7 rendered — same matrix the /media preview surfaces for this asset.

Composition layer × scene 10 scenes · 235.4s · comp_id + rendered still + tier + the script

#Layer (comp_id · still · tier)BeatTimecodeMotionLogoAudioVO / on-screen / caption
1s1
matches intent
shared field
signature-3d
open0–31.6sspatial-parallaxicon·liquid-chrome♪ bed_in
This week the calendar says one thing. Access into advantage. Most people think that means knowing the right person at a party. It does not. Access is just a door. Advantage is what you build on the other side of it. We run The Big T-M as a studio, in the open, with AI doing the work. Today we open the part most agencies hide, the relations machine, and show you how a model turns a cold world into a warm one. No hype, just the moves.
on-screen: Access into advantage, in public
expected on screen: red ground · tetra hero in the shared Signal Field · Nexus leads · node-graph · spatial-parallax · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape tetraground redtreatment liquid-chromemotion spatial-parallaxpower summoninstrument summon→Access into advantage, in public
2s2
matches intent
NumberedList
template
teach31.6–60.0skinetic-buildicon·white-knockout♪ node_lock
Start with a list, not a pitch. Open a coding agent in the terminal, Claude Code, and point it at a tool that scrapes the web. We use Apify through an MCP server, a small adapter that lets the agent run a scraper directly. Ask it to pull every reporter, podcast host, and event organizer in your space, with their last five posts. You now have a contact graph. The model reads it, you do not have to.
on-screen: Build the contact graph first
expected on screen: red ground · a NumberedList panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id NumberedListvisual node-graphshape tetraground redtreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Build the contact graph firstcurate items, nodes
3s3
matches intent
ChecklistCard
template
teach60.0–83.8skinetic-buildicon·liquid-chrome♪ node_lock
A press list is useless if you pitch everyone the same line. Feed those last five posts back to the model and ask one question. What does this person actually cover. The agent sorts your list by topic, so the energy reporter gets your energy story and nobody gets a story they would never run. This is the publicist move, done by reading, not guessing.
on-screen: Match the angle to the human
expected on screen: red ground · a ChecklistCard panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id ChecklistCardvisual node-graphshape tetraground redtreatment liquid-chromemotion kinetic-buildpower morphinstrument morph+laser→Match the angle to the humancurate items, nodes
4s4
matches intent
NodeGraphCard
template
teach83.8–106.1skinetic-buildicon·white-knockout♪ node_lock
Now draft. Give the model the reporter's own headlines as a style sample and tell it to write a three sentence pitch in that register. Short. One hook, one fact, one ask. The model matches their tone because it just read their work. You read the draft, cut a line, and it sounds like a person, not a press release blast.
on-screen: Write the pitch in their words
expected on screen: red ground · a NodeGraphCard panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id NodeGraphCardvisual node-graphshape tetraground redtreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Write the pitch in their wordscurate hub, nodes
5s5
matches intent
CheatSheet
template
teach106.1–126.3skinetic-buildicon·liquid-chrome♪ node_lock
Do not buy a sending tool. Wire the agent into Gmail through its MCP and let it create drafts, one per reporter, in your real account. You open the folder, glance at each, and hit send yourself. The relationship stays human. The typing, the sorting, and the follow-up reminders are the part the machine carries.
on-screen: Send from your own inbox
expected on screen: red ground · a CheatSheet panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id CheatSheetvisual node-graphshape tetraground redtreatment liquid-chromemotion kinetic-buildpower morphinstrument morph+laser→Send from your own inboxcurate nodes, points, uses
6s6
matches intent
StepFlow
template
teach126.3–147.2skinetic-buildicon·white-knockout♪ node_lock
Partnerships are the same engine pointed sideways. Ask the model to scan companies that sell to your buyer but do not compete with you. The accountant who serves the same shops you do. Now you are not begging for a favor, you are offering their list something useful. The agent finds the overlap. You make the call.
on-screen: Find the partner, not the logo
expected on screen: red ground · a StepFlow panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id StepFlowvisual node-graphshape tetraground redtreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Find the partner, not the logocurate nodes, steps
7s7
matches intent
LogStream
template
teach147.2–168.1skinetic-buildicon·liquid-chrome♪ node_lock
Going to a conference. Before you fly, have the agent pull the attendee and speaker roster from the event page and cross it against your contact graph. It hands you ten names worth your time and a one line reason for each. You walk in with a plan instead of a name tag and a free drink.
on-screen: Turn an event into a list
expected on screen: red ground · a LogStream panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id LogStreamvisual node-graphshape tetraground redtreatment liquid-chromemotion kinetic-buildpower morphinstrument morph+laser→Turn an event into a listcurate nodes, rows
8s8
matches intent
JsonDiff
template
teach168.1–189.7skinetic-buildicon·white-knockout♪ node_lock
None of this matters if you forget who you talked to. Point the agent at Airtable through its MCP and have it log each pitch, reply, and meeting as a row. Now your relations are a database you can ask questions of, not a memory you will lose. Who went quiet. Who said maybe. The model keeps the ledger.
on-screen: Log every touch in one place
expected on screen: red ground · a JsonDiff panel over a dimmed Signal Field · Nexus leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id JsonDiffvisual node-graphshape tetraground redtreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Log every touch in one placecurate fileName, lines, nodes
9s9
matches intent
KpiGrid
template
proof189.7–210.2sreceipts-counticon·liquid-chrome♪ node_lock
Here is the honest part. We built this whole relations system by pointing it at our own studio first. The contact graph, the drafts, the Airtable log, all live, all built from one source file that also builds this video. We are not selling you a method we have never run. You are watching it run.
on-screen: We ran this on ourselves
expected on screen: red ground · a KpiGrid panel over a dimmed Signal Field · Nexus leads · receipts · receipts-count · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id KpiGridvisual receiptsshape tetraground redtreatment liquid-chromemotion receipts-countpower receiptsinstrument spotlight→We ran this on ourselvescurate kpis, statsLabels
10s10
matches intent
shared field
signature-3d
resolve210.2–235.39999999999998scoalescenceicon·liquid-chrome♪ bed_out
So that is access into advantage. Find the people with a scraper. Sort them by what they cover. Write in their voice. Send from your own hands. Track it all in one table. The door is just step one. The build is the work. See the whole studio operating, in public, at office dot temerarii dot xyz, and then go point one of these moves at your own list.
on-screen: See the machine at office.temerarii.xyz
expected on screen: red ground · tetra hero in the shared Signal Field · Nexus leads · coalescence · coalescence · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape tetraground redtreatment liquid-chromemotion coalescencepower coalescenceinstrument coalescence→See the machine at office.temerarii.xyz

Format stack 1 aspects · same scenes[], re-cropped

16:9
1920×1080
X/Twitter · YouTube · LinkedIn video

Channels 2 destinations

YouTubeBlog

Social captions supplemental published copy · per channel (comp_id level)

youtubeTurn Access Into Advantage: A Relations Machine That Finds, Sorts, and Pitches Most people think access means knowing the right person at a party. It does not. Access is just a door. Advantage is what you build on the other side. In this video we open the part most agencies hide, the relations machine, and show how a model turns a cold world into a warm one. The moves: - Build the contact graph first. Open Claude Code in the terminal and point it at a scraper, we use Apify through an MCP server. Ask it to pull every reporter, podcast host, and event organizer in your space with their last five posts. You now have a contact graph. - Match the angle to the human. Feed those posts back to the model and ask what each person actually covers. The agent sorts your list by topic so the right story reaches the right person. - Write the pitch in their words. Give the model the reporter's own headlines as a style sample and have it write a three-sentence pitch in that register. One hook, one fact, one ask. - Send from your own inbox. Wire the agent into Gmail and let it create drafts in your real account. You glance at each and hit send yourself. The relationship stays human. - Find the partner, not the logo. Ask the model to scan companies that sell to your buyer but do not compete with you, like the accountant who serves the same shops. - Turn an event into a list. Have the agent pull the attendee roster and cross it against your contact graph, then hand you ten names worth your time. - Log every touch in one place. Point the agent at Airtable through its MCP and have it log each pitch, reply, and meeting as a row you can ask questions of. The honest part: we built this whole system by pointing it at our own studio first. The contact graph, the drafts, the Airtable log, all live, all from one source file. You are watching it run at office.temerarii.xyz. The takeaway: find people with a scraper, sort by what they cover, write in their voice, send from your own hands, track it all in one table. Keywords: media relations, Apify, Gmail, Airtable, MCP, outreach.

Cross-links every lens is a view on this one record