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-W51-Satkind longformweek W51date 2026-12-26campaign longform-youtubepillar it_devbeat asset videoduration 163.2sground blackscenes 8

Checklist the per-video bar — engine/sim

98.4/100
plain languagevo coverageno dead airuniquenesscaption fitcompletenesscleanliness
quantitative quality · weights learn from your reviews (engine.sim.memory review longform-W51-Sat good|bad)
⚠ 1 flag(s) — not yet ship-ready: copy_generic · 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 · 163.2s · 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
The week is about training teams on AI for go-to-market, and today is the brand slice done deep: teaching a model to write like you, so your content stays yours at scale. A generic model writes generic words. By the end of this you will know how to turn your brand voice into a rule a model follows, instead of a feeling only one person can do.
on-screen: Teach the model your voice
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→Teach the model your voice
2s2
matches intent
NumberedList
template
teach24.5–45.4skinetic-buildicon·wireframe♪ node_lock
First move: gather the lines you are proud of. Real captions, real emails, the ones that sounded like you. Put twenty of them in one file. This is your voice, on paper, in examples. The team learns that voice is not a mystery, it is a stack of sentences you can point at and say, like this.
on-screen: Collect your real best lines
expected on screen: black ground · a NumberedList panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id NumberedListvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Collect your real best linescurate items, nodes
3s3
matches intent
ChecklistCard
template
teach45.4–66.7skinetic-buildicon·wireframe♪ node_lock
Second move: write the other half of the rule, the words and moves you ban. The empty hype words everyone reaches for, the fake urgency, the exclamation marks. Give the model the no-list as clearly as the yes-list. The team learns that a strong voice is defined as much by what it refuses as by what it says.
on-screen: Name the words you never use
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→Name the words you never usecurate items, nodes
4s4
matches intent
NodeGraphCard
template
teach66.7–88.30000000000001skinetic-buildicon·wireframe♪ node_lock
Third move: feed the model your file and ask it to write one new caption in that voice, then put it next to a real one. If a teammate cannot tell which is which, the rule is working. If they can, you add the missing rule and try again. This side-by-side check is how the team tightens the voice.
on-screen: Make it match a sample
expected on screen: black ground · a NodeGraphCard panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id NodeGraphCardvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Make it match a samplecurate hub, nodes
5s5
matches intent
AnnotatedDiagram
template
teach88.3–108.1skinetic-buildicon·wireframe♪ node_lock
Fourth move: once the voice holds, point it at every place the brand shows up, the same rule on the website, the email, the social post. One voice file, many surfaces. The team learns that consistency is not extra work, it is the same rule reused, so the brand sounds like one person everywhere.
on-screen: Apply it across every surface
expected on screen: black ground · a AnnotatedDiagram panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id AnnotatedDiagramvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→Apply it across every surfacecurate callouts, nodes
6s6
matches intent
TerminalRun
template
teach108.1–127.19999999999999skinetic-buildicon·wireframe♪ node_lock
Honest proof: every line you have heard this week came from a model following our written voice file, checked against our real lines. No claim of perfection and no invented score. The proof is that the brand has sounded like one steady voice all week, off one source file and one rulebook.
on-screen: This week is the side-by-side test
expected on screen: black ground · a TerminalRun panel over a dimmed Signal Field · Faber leads · node-graph · kinetic-build · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id TerminalRunvisual node-graphshape boxground blacktreatment wireframemotion kinetic-buildpower morphinstrument morph+laser→This week is the side-by-side testcurate nodes
7s7
matches intent
ComparisonTable
template
proof127.2–145.2sreceipts-counticon·wireframe♪ node_lock
Wait, one caution. A voice file is not set and forget. When the brand grows, the team has to feed it new winning lines and retire old rules, or the voice goes stale. Treat it as a living file your people keep tending. That tending is the real skill.
on-screen: Run the brand-voice loop
expected on screen: black ground · a ComparisonTable panel over a dimmed Signal Field · Faber leads · receipts · receipts-count · icon·wireframe logo · caption bottom-left
spec (the prompt): comp_id ComparisonTablevisual receiptsshape boxground blacktreatment wireframemotion receipts-countpower receiptsinstrument spotlight→Run the brand-voice loopcurate colA, colB, rows, statsLabels
8s8
matches intent
shared field
signature-3d
resolve145.2–163.2scoalescenceicon·white-knockout♪ bed_out
So the takeaway: collect your best lines, name the words you refuse, check the model side-by-side, and reuse one voice file everywhere. Now your brand voice is a thing your team can run, not a thing trapped in one head. Watch it run on live brand work at office.temerarii.xyz.
on-screen: Make your voice repeatable
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→Make your voice repeatable

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)

youtubeTeach a model your brand voice so your content stays yours at scale A generic model writes generic words. By the end of this video you will know how to turn your brand voice into a rule a model follows, instead of a feeling only one person can do. The moves: - Collect your real best lines. Gather the lines you are proud of, real captions and emails, the ones that sounded like you. Put twenty of them in one file. That is your voice, on paper, in examples. Voice is not a mystery; it is a stack of sentences you can point at and say, like this. - Name the words you never use. Write the other half of the rule: the words and moves you ban. The empty hype words everyone reaches for, the fake urgency, the exclamation marks. Give the model the no-list as clearly as the yes-list. A strong voice is defined as much by what it refuses. - Make it match a sample. Feed the model your file and ask it to write one new caption in that voice, then put it next to a real one. If a teammate cannot tell which is which, the rule is working. If they can, add the missing rule and try again. - Apply it across every surface. Once the voice holds, point it at every place the brand shows up: the website, the email, the social post. One voice file, many surfaces. Consistency is not extra work; it is the same rule reused. One caution: a voice file is not set and forget. As the brand grows, feed it new winning lines and retire old rules, or the voice goes stale. Treat it as a living file your people keep tending. The honest proof: every line you have heard this week came from a model following our written voice file, checked against our real lines. No claim of perfection and no invented score. The proof is that the brand has sounded like one steady voice all week. Watch it run on live brand work at office.temerarii.xyz Keywords: brand voice with AI, voice file, banned words list, side-by-side voice check, consistent writing at scale, AI copy on brand

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