Tools / Analysis
Selendis wants Archon to learn how you cut, not just what you prompt.
The team promises edits from raw footage in minutes, with preferences learned from your catalogue, references and changes. Early access is through a waitlist; the learning mechanism remains largely unexplained.
An editor’s taste lives in decisions: which take survives, where a pause ends, how long a face stays on screen. Selendis is proposing to make those decisions part of an AI editor’s input. On 3 November, the company introduced Archon, which it describes as a video foundation model for turning raw footage into an edit in minutes.
The more interesting promise is what happens after that first edit. Selendis says the editor adapts to an individual creator or team through their corrections, previous videos and chosen references. If that works reliably, a correction could become guidance for the next project rather than an instruction you have to repeat.
Three ways to show an editor what you mean
The announcement names three sources of preference: decisions made inside Selendis’s collaborative editor, the creator’s existing video catalogue, and reference videos. These are different kinds of evidence. A finished film shows the result; an edit inside the timeline can show what the author chose to change. A reference points toward a desired rhythm or treatment without necessarily saying which part matters.
Selendis calls the initial edit zero-shot. Its public description does not explain whether subsequent personalisation changes model weights, stores preferences, retrieves past decisions or combines several methods. Nor does it specify how it distinguishes a lasting preference from a one-off choice for a particular client. Calling the system a foundation model does not settle those questions.
Sources: Selendis / X [1] · Selendis [2]
Archon’s announcement and product demonstration
Loading post from X…
The company’s selected demonstration. Its speed and personalisation claims have not been independently tested here.
An announcement, with access still gated
The website describes Selendis as being in early access and directs visitors to a waitlist. The launch post names Steve Aoki and industry specialists among the people behind it. Neither source provides a public benchmark, a breakdown of failed edits or enough timing detail to turn “minutes” into a predictable production budget.
For an editor, the useful trial would start with familiar rushes and a finished cut to compare against. Can Archon retain the intended story, accept a precise revision, and carry that preference into the next job without applying it indiscriminately? Those are proposed tests, not results we obtained. They measure the value of personalisation more directly than a polished launch reel.
Sources: Selendis [2] · Selendis / X [1]
SOURCES & METHOD
The record behind the article
Based on Selendis’s announcement and product page. We have not tested the editor. The cover is a conceptual editorial illustration, not a product screenshot.
- Introducing Archon Selendis / X · 3 November 2026
- Personal AI Video Editors — product and early-access information Selendis · Checked 4 November 2026
