Nicky Saunders hands the same clip to two AI editors — one built on templates, one built on a single sentence — and only one of them invents a random uncle.
AI video editors now split into two real workflows — pick-a-template speed versus describe-it-in-a-sentence control — and choosing between them is a workflow-fit decision, not a quality contest.
Nicky Saunders feeds the same raw clip to two AI video editors: Submagic, a template-first auto-editor, and AI Producer, Opus Clip's new beta that you edit by typing a single instruction instead of touching a timeline. Submagic's pick-a-style flow produces a fast, well-captioned cut, but its automatic b-roll matching drops in an unrelated stranger during an emotional line — fixed with a two-click asset swap. AI Producer skips template selection entirely: given one sentence ('polish this talking head, add motion graphics where you see fit'), it asks a clarifying question about how aggressively to clean the delivery, then produces a comparable cut without the b-roll error. The conclusion isn't that one tool wins — it's that template-pickers and talk-it-out people need different tools, and both lanes now exist.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →
Nicky says she didn't edit this video herself and frames the test: the same raw clip, given to two AI editors, one template-based and one prompt-based.
Uploads the clip to Submagic, picks a persona-style template (creative/energetic) from a gallery of preset looks, and lets the auto video editor process it.
Submagic's auto-generated cut inserts an unrelated older man as reaction b-roll over the line 'you felt safe enough to cry' — Nicky swaps the asset in two clicks, then removes a long silence and the bad line entirely rather than patch around it.
Rates the auto-edit — no manual edits needed beyond the b-roll swap and silence removal — and singles out Submagic's caption styles as best-in-class; names it as the right tool for creators who just want to pick a template.
Uploads the same clip into Opus Clip's AI Producer beta, selects a 'smart model' (Claude, GPT options available), and sets it to ask clarifying questions instead of auto-running straight through.
Types a single instruction — polish this talking head, add motion graphics where fitting — and answers the tool's follow-up question about how aggressively to clean the delivery ('standard clean').
AI Producer returns a cut from the one-sentence brief with no timeline editing and no random-stranger b-roll error; Nicky compares it directly against the Submagic version.
Reframes the comparison as a persona question — template-picker versus talk-it-out person — rather than a which-tool-wins verdict, and asks viewers to comment which lane they're in.
AI auto-editing has split into two real workflows — pick a template and accept the tool's judgment, or brief the edit in one sentence — and the right choice depends on how you like to work, not which tool is objectively better.
“Whose man was that?”
“That uncle... gone. Two clicks, no need to manipulate the timeline.”
“What I will say for Submagic, their caption styles are undefeated.”
“It's not about which tool is better. It's just what type of person are you.”
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
Nicky Saunders opens with a flat claim — she didn't edit this video — before revealing she handed the same raw clip to two competing AI editors, one of which returned a random man she's never met standing in for an emotional beat.
The video's core framing: AI video editors now split into two interaction models, and the right one depends on how the creator prefers to work, not which tool edits better.
“I also want you to check underneath this video to see if you've subscribed to this channel. If you haven't, please do because this helps out the channel so we can test out more of these tools so you don't have to waste your money.”
Soft, low-key ask stacked after the payoff/comparison rather than up front; paired with a comment prompt (template lane vs. talk-it-out lane) and a next-video pointer.
00:00
00:10
00:1900:23
00:30
00:41
00:49
00:56
01:04
01:11
01:19
01:26
01:3201:42
01:46
01:57
02:05
02:11
02:19
02:27
02:35
02:43
02:49
02:57
03:05
03:12
03:20
03:27
03:35
03:43
03:50
03:58
04:05
04:13
04:20
04:28
04:36
04:43
04:51
04:58
05:06
05:1305:21
05:28
05:36
05:44
05:48
05:59
06:06
06:14
06:18
06:2806:37
06:44
06:52
06:59
07:07
07:14
07:2207:29
07:37
07:45
07:52
08:00
08:07
08:15
08:22
08:3008:41
08:47
08:53
09:00
09:08
09:1209:21
09:32
09:38
09:43
09:5309:58
11:20A creator talks to ChatGPT's new voice mode like a cohost, turning a two-minute ramble into hooks, a thumbnail prompt, and a finished image — then the voice briefly glitches into what sounds like a different person.
July 11th
13:02A creator with zero web design experience walks through the exact prompts, tools, and 5-step loop she used to get Claude Fable 5 to build a fully animated, subway-themed personal website.
July 5th
23:34A 23-minute end-to-end guide to making AI-edited video from natural language -- no timeline, no drag-and-drop, just HTML rendered locally.
June 23rd
11:36An 11-minute field report on three AI tools that survived a week of real production work.
June 12th
11:47A creator with 20 years of editing experience built a custom Claude skill that edits a 40-minute raw talking-head video in 12 minutes -- then gave it away free.
June 15th
05:48A 5-minute walkthrough of Higgsfield -- the AI plugin that puts color grading, object removal, background removal, video generation, and 8K upscaling inside DaVinci Resolve.
June 12th