Approved scenes
Created once (increasingly generated with AI), reviewed by your team, then reused as building blocks.
Glossary · Video assembly
What is video assembly?
Video assembly is how personalized video scales: instead of producing a new video per customer by hand or by prompt, a system composes each video at runtime from approved scenes, filling personalized elements inside them (on-screen text, imagery, voiceover) from customer data. One video per customer per event, and the same inputs always produce the same video.
Assembly answers a question generation cannot: how does one message become a million correct videos? The answer is to stop treating each video as a file to produce. Scenes are created once (filmed, designed or generated with AI, under your review) and become reusable building blocks. When a customer reaches a moment (a renewal, a first bill, a claim update), rules read that customer's data and pick which scenes they see, in which order and language, and the video is composed on the spot.
Selection is only half of it. Inside each scene, personalized elements are filled from the customer's own record: on-screen text shows their name and numbers, imagery adapts to their product or city, and the voiceover speaks their details in their language. Two customers can share the same scene and still each see their own video. The result is one video per customer per event, not a template with a name pasted on.
The properties enterprises buy follow from this design. Change one scene and every video assembled from it updates; there is no archive of a million files to regenerate. And because assembly follows rules rather than a prompt, it is deterministic: the same record and the same rules always produce the same video. That is what makes per-customer video workable in regulated industries, and it is the job Lont's patent-pending engine does in real time.
Key points
How assembly works
Rules select
Customer data decides which scenes each person sees, in which order and language, per event.
Elements personalize
Inside each scene, text, imagery and voice are filled from the customer's record.
Deterministic output
Same record, same rules, same video, every time, at any volume.
FAQ
Frequently asked questions
How is assembly different from AI video generation?
Generation creates content: a prompt becomes a clip, and a million different clips means a million generations, each one new and each one to check. Assembly composes from content that already passed review, so volume adds no new risk and little new cost. In practice they stack: generative AI helps create the scenes, assembly turns them into each customer's video.
What can be personalized inside a scene?
On-screen text (names, amounts, dates), imagery (product, brand, location), the spoken voiceover, subtitles and language, and the next step the video offers. Anything your rules can read from your systems can appear in the video, and a journey reads only the fields it actually uses.
Does every video still need approval?
You approve scenes, not videos: every video assembled from an approved scene inherits that approval, which is what makes a million-video journey reviewable at all. How strict you make it stays your choice. Most regulated teams require sign-off on every scene; you can also create AI-generated scene variants for parts where compliance matters less. The system enforces whatever you set.
Why does assembly matter for compliance?
Because it turns an impossible review problem into a finite one. Nobody can review a million individual videos; a scene library can be reviewed in an afternoon. Compliance itself stays your responsibility and your call: what assembly provides is that the content your team approved is what customers see, reproducibly.
Get started
See one message assembled per customer.
Bring a message you send at scale and we'll show it assembled live: same scenes, different customers, each with their own video.
- −10.9% churn
- 13 → 36 NPS
- 41–52% click-to-open
Thanks, that's in.
We'll be in touch within one business day to set up your walkthrough.