150 clips from your film. Posted daily. Then more of whatever strangers actually watch.

You tell it what the film is for. It reads the entire film first — transcript, scenes, the shape of the story — and every clip arrives with the reason it was chosen.

From $1,000 a title, then $250 a month. Three accounts included.

01Proof

The same library. The same schedule. Two very different results.

One filmmaker ran two pages on the same film library at the same cadence — two posts a day, near-identical captions, near-identical on-screen text. One page was fed by clips a human editor cut. The other was fed by ours.

Ours caught fire.
The other one is still waiting.

Same library, same cadence, same captions. The only variable was which clips went out.

Head-to-head, two pagesTwo posts a dayOne variable

“The software really found the tense moments in a slow-burn heavy drama, and leaned in hard into what is visually and emotionally appealing.”

Actor-filmmaker, five features

“The clips are fantastic. Really wonderful. You guys nailed it.”

On a 90-clip delivery, after one round of notes

37 of 90 clips sent back to be re-cut by our own review, before the filmmaker saw any of them.

The sound-off pass, last full batch

02Why this exists

Your distributor stopped marketing your film the week it launched.

That is not a complaint about distributors. It is how the model works: they ramp for release, then move to the next title. What is left is a finished film, a shrinking audience, and no budget.

Clips are the only marketing you can afford to run forever. The problem has never been whether to cut them. It is that cutting them, posting them twice a day, watching for flags, and working out what landed is a permanent job nobody hired for.

So we do all four parts.

03The loop

Four stages. The fourth one is why the first three keep getting better.

The ring closes. That is the part you are buying.

  1. 01

    It reads the film before it cuts it

    Transcript, scene structure, the whole arc, in one pass before a single clip is chosen. That is why it finds the tense beat in a quiet scene instead of the loudest moment in a fast one.

    You set the goal for the titleFinal act excluded by default

  2. 02

    It cuts about 150, and every one carries its reason

    Not a score. A sentence saying what the moment is doing and who it should catch. Disagree in plain language and the next batch reflects it.

    Square and verticalFull source qualityCaption files suppliedChecked on a phone, sound off

  3. 03

    It posts to channels you own

    Two a day per account, up to three accounts, and nothing goes live until a human approves the queue. Your pages, not rented ones, so the audience you build is an asset you keep.

    Flags caught and repostedA log of what went where

  4. 04

    It reads what happened, and cuts again

    Winners rank on cold-audience watch time: strangers with no reason to be generous, staying anyway. Near a dozen of the 150 clearly hold, and those come back as new variants.

    Underperformers drop outBank recycles at about four months

Stage four feeds stage one. Your second film starts where the first one finished.

The reasons behind the winners become the brief for the next title in your catalogue.

~150

clips per title, square and vertical

5 days

typical turnaround from a complete intake

2 a day

posted per account, human-approved queue

4 months

before the bank recycles to a fresh feed

04The output

Every clip tells you why it was chosen.

This is how a batch arrives — every clip playing in the page, its opening line underneath, its reason beside it, filterable by length. Scroll it the way you would if it were yours.

batch-01 · clip wall Sample layout 12 of 12 shown
9:160:14Re-cut

“He knew before she said it.”

Quiet two-hander. Cuts one line before the answer.

9:160:22

“Nobody drives out here at night.”

Threat with nothing violent on screen. Safe on every platform.

1:10:08

“Say it again.”

Shortest usable beat in the film. Built for the repost slot.

9:160:31

“The money was never the point.”

Second-act turn, said out loud. Lands with no context.

9:160:17

“You brought him here.”

Face-forward accusation. Reads with the sound off.

1:10:45

“Walk me through it one more time.”

Longest hold in the batch. Kept for people who already follow.

9:160:11

“Don’t.”

One word, one cut. Tests whether the hook alone holds the scroll.

9:160:26Re-cut

“That’s not what the file says.”

Procedural beat, for the crowd that watches this genre for the mechanics.

1:10:19

“I told you what she was.”

Cast-forward. The recognisable face is in frame throughout.

9:160:38

“There’s a version of this where we both walk away.”

Slow build, no spoiler. Ends before the scene resolves.

9:160:13

“Look at me.”

Pure performance. Carries no plot, so it runs at any point in the rotation.

1:10:55

“Everyone in this room has lied to me.”

Ensemble wide. The one clip that shows the whole cast.

Frames and lines here are placeholders. A live wall holds about 150 clips off one feature and ships as its own page you can send to anyone.

The rationale card

One clip, and the reason it was cut — written before anyone knew how it would do.

0:14

Quiet two-hander. Cuts one line before the answer, so the scroll has to stop to find out.

Then it is matched against what a cold audience did with it. That match is what picks the next batch, and the one after that.

The winner’s children

One clip that worked, and the five variants cut from it afterwards. Different entry points, different lengths, same moment. Stage four, in one picture.

0:14
0:08
0:11
0:17
0:22
0:29

05Reporting

You will not be given a dashboard to learn.

Every tool in this category sells you a reporting interface and calls it a feature. You did not make a film so you could study charts.

Once a week you get four things: what went out, what a cold audience actually sat through, which clips are being cut again because they worked, and which ones we stopped posting.

The underlying data is there if you want it. Most people want the answer.

06Pricing

One build per title. Then it runs.

Three social accounts included per title.

1 to 5 titles

$1,000

per title, once — to build the bank

$250

per title, per month — to keep it working

6 to 20 titles

$850

per title, once

$225

per title, per month

21 or more

Let’s talk

Catalogue pricing, built around how many titles are actually earning.

What the build includes

  • About 150 clips, single scene, at full source quality
  • Square and vertical, routed per platform
  • An on-screen text hook on every clip
  • Caption files supplied separately
  • Music stripped where rights are a risk
  • Two rounds of re-cuts against your notes
  • Typically five business days from a complete intake

What the monthly includes

  • Scheduling and posting, up to two per account per day across three accounts
  • Nothing live until a human approves the queue
  • Flag and shadowban monitoring, with reposts
  • Weekly numbers off every account
  • Winners ranked on cold-audience watch time and re-cut for reposting
  • Four-month recycling of the back catalogue
  • Instagram linked to Facebook so view counts combine
  • Every clip tagged with why it was chosen, matched against what performed
  • A/B testing of hooks, captions, lengths and formats
  • A monthly call

What it does not include

  • Social accounts beyond three
  • Your own scheduler seat
  • Storage and licensed music
  • Paid promotion or ad spend
  • New titles — each is its own build plus its own monthly line

One title runs $4,000 in the first year and $3,000 a year after. If you are sitting on six or more, the per-title price drops and the loop runs across all of them at once — which is the version of this that actually moves a library.

07Questions

What usually stops people.

“AI clips look like AI clips.”

Most do, because most tools look for spikes in the audio and cut around them. This one reads the script and the scene structure first, then cuts to a goal you set. The wall above is unedited output. Judge it there rather than here.

“My film is slow. There’s nothing to clip.”

That was the harder of our two test cases and it produced the better result. Tension in a quiet scene is a pattern in the material, not a volume level.

“I’ve paid for this before and got nothing.”

So had the filmmaker above — five figures, one title, posted to accounts he did not own, and nothing at all once the launch window closed. Three things are different here. The channels are yours. The posting does not stop. And what worked this month decides what goes out next month.

“Will it spoil my movie?”

No. The final act is excluded by default. Set a different cutoff if you want one.

“What about music and rights flags?”

Score is stripped where rights are a risk, and the posting layer watches for flags and reposts what gets caught. No one can promise a platform will behave, and we will not pretend otherwise.

“What happens when I run out of clips?”

You do not, really. Winners get re-cut into new variants, and the bank recycles after about four months to an audience that has scrolled a long way past it. That is the difference between a campaign and a catalogue.

“How much of my time does this take?”

Fifteen minutes to brief a title. An hour or so to review a batch. After that, a monthly call if you want one.

“I don’t want another dashboard.”

You are not getting one. Four lines a week, and the decisions already made.

Your film is finished. The marketing isn’t.

Send the file and tell us what you want it to do. About 150 clips come back, each with the reason it was chosen. Then it starts posting, and it starts learning.

From $1,000 a title, then $250 a month. Two rounds of re-cuts included.

Book time directly: calendar.app.google · team@nexrizen.com · nexrizen.com