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-W50-Monkind longformweek W50date 2026-12-14campaign longform-youtubepillar it_devbeat —asset videoduration 157.7sground blackscenes 8

Checklist the per-video bar — engine/sim

95.1/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W50-Mon good|bad)
⚠ 3 flag(s) — not yet ship-ready: generic_scenecustom_element_dup · see docs/strategy/VIDEO-CHECKLIST.md

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 8 scenes · 157.7s · comp_id + rendered still + tier + the script

#Layer (comp_id · still · tier)BeatTimecodeMotionLogoAudioVO / on-screen / caption
1s1
matches intent
shared field
signature-3d		open0–24.5sspatial-parallaxicon·white-knockout♪ bed_in
Today is growth and performance marketing, and we learn it by running a real test, not reading about funnels. By the end your team will have launched one small experiment, measured it, and known whether it worked. The promise of this week is teams that ship real value, so we are going to ship something today. Open your laptops. We start with the one number that matters.
on-screen: Growth, taught hands-on
expected on screen: black ground · box hero in the shared Signal Field · Faber leads · node-graph · spatial-parallax · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual node-graphshape boxground blacktreatment white-knockoutmotion spatial-parallaxpower summoninstrument summon→Growth, taught hands-on
2s2
matches intent
NumberedList
template		teach24.5–50.4skinetic-buildicon·ember-fill♪ node_lock
First move: choose a single number to move this week. Not ten. One. Maybe it is replies to a cold email, or clicks on one ad. We have the team write it on the board and define exactly how it is counted. Most growth work fails because nobody agreed on the target. The lab is each person picking their one metric and saying out loud how they will know it changed.
on-screen: Pick one metric, ignore the rest
expected on screen: black ground · a NumberedList panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·ember-fill logo · caption bottom-left
spec (the prompt): comp_id NumberedListvisual node-graphshape boxground blacktreatment ember-fillmotion kinetic-buildpower morphinstrument morph+laser→Pick one metric, ignore the restcurate items, nodes
3s3
matches intent
ChecklistCard
template		teach50.4–74.9skinetic-buildicon·wireframe♪ node_lock
Second move: build the test. We take one ad or email and make two honest versions, A and B, that differ in just one thing. To go faster, we have the agent draft ten variations of the headline in plain language, then the team picks the two strongest by hand. The model gives you options. You give it taste. That split is the skill we are teaching.
on-screen: Write two versions, let AI vary them
expected on screen: black ground · a ChecklistCard panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ChecklistCardvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Write two versions, let AI vary themcurate items, nodes
4s4
matches intent
JsonDiff
template		teach74.9–99.4skinetic-buildicon·white-knockout♪ node_lock
Third move: launch it small and watch. We set a tiny budget or a small send, so a wrong guess costs little. Then we connect the ad platform or email tool through an MCP server so the agent can pull results as they come in. The team asks, which version is winning, in words, and reads the answer together. Real numbers, same hour, on their own screen.
on-screen: Ship small, watch it live
expected on screen: black ground · a JsonDiff panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id JsonDiffvisual node-graphshape boxground blacktreatment white-knockoutmotion kinetic-buildpower morphinstrument morph+laser→Ship small, watch it livecurate fileName, lines, nodes
5s5
matches intent
WireframeMock
template		teach99.4–123.9skinetic-buildicon·liquid-chrome♪ node_lock
Here is honest proof. We do not invent a percentage. In our own labs, the small test a team runs on day one usually finds one clear winner they would have guessed wrong. That single fix often covers the cost of the day. We have watched teams kill an ad they loved because the boring version quietly won. The data, not the loudest opinion, made the call.
on-screen: It paid for the room
expected on screen: black ground · a WireframeMock panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·liquid-chrome logo · caption bottom-left
spec (the prompt): comp_id WireframeMockvisual node-graphshape boxground blacktreatment liquid-chromemotion kinetic-buildpower morphinstrument morph+laser→It paid for the roomcurate nodes
6s6
matches intent
StackTrace
template		teach123.9–147.70000000000002skinetic-buildicon·ember-fill♪ node_lock
The takeaway: growth is not a big bet, it is a steady habit of small honest tests where the number decides. Keep what wins, drop what loses, run the next one. Your team now has the loop and the tools to repeat it without us. See the full system we run on ourselves at office dot temerarii dot xyz, and bring a real campaign tomorrow.
on-screen: Keep the winner, drop the rest
expected on screen: black ground · a StackTrace panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·ember-fill logo · caption bottom-left
spec (the prompt): comp_id StackTracevisual node-graphshape boxground blacktreatment ember-fillmotion kinetic-buildpower morphinstrument morph+laser→Keep the winner, drop the restcurate errMsg, errType, fix, frames, nodes
7s7
first render · fix pending
RankList
templategeneric_scenecustom_element_dup		proof147.7–152.7sreceipts-counticon·wireframe♪ node_lock
Custom apps + dashboards: own the code and the IP
on-screen: Chapter 6
expected on screen: black ground · a RankList panel over a dimmed Signal Field · Faber leads · receipts · receipts-count · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id RankListvisual receiptsshape boxground blacktreatment wireframemotion receipts-countpower receiptsinstrument spotlight→Chapter 6curate rows, statsLabels
8s8
first render · fix pending
shared field
signature-3dgeneric_scene		resolve152.7–157.7scoalescenceicon·white-knockout♪ bed_out
Custom apps + dashboards: own the code and the IP
on-screen: Chapter 7
expected on screen: black ground · box hero in the shared Signal Field · Faber leads · coalescence · coalescence · icon·white-knockout logo · caption bottom-left
spec (the prompt): comp_id shared Signal Fieldvisual coalescenceshape boxground blacktreatment white-knockoutmotion coalescencepower coalescenceinstrument coalescence→Chapter 7

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)

youtubeGrowth marketing taught by running one real A/B test today We learn growth by running a real test, not reading about funnels. By the end of this video your team will have launched one small experiment, measured it, and known whether it worked. The moves: - Pick one metric, ignore the rest. Choose a single number to move this week, not ten. Maybe replies to a cold email, or clicks on one ad. Write it on the board and define exactly how it is counted. Most growth work fails because nobody agreed on the target. - Write two versions, let AI vary them. Take one ad or email and make two honest versions, A and B, that differ in just one thing. To go faster, have the agent draft ten headline variations in plain language, then pick the two strongest by hand. The model gives you options; you give it taste. - Ship small, watch it live. Set a tiny budget or a small send, so a wrong guess costs little. Connect the ad platform or email tool through an MCP server so the agent can pull results as they come in. Ask which version is winning, in words, and read the answer together. - Keep the winner, drop the rest. Growth is a steady habit of small honest tests where the number decides, not the loudest opinion. Keep what wins, drop what loses, run the next one. The honest part: in our own labs, the small test a team runs on day one usually finds one clear winner they would have guessed wrong. The data made the call, not the loudest voice. No invented percentage, just the method. See the full system we run on ourselves at office.temerarii.xyz Keywords: growth marketing, A/B testing, one metric, AI headline variations, small budget test, performance marketing for teams

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

→ Wireframe · scenes→ Calendar week→ Media (audience preview)→ Storyboard hub→ Day view→ Channel view