A YouTuber hands 24 minutes of raw, mistake-filled footage to ChatGPT connected straight into DaVinci Resolve's new MCP server, and watches it cut, retime, brand, and export the finished video almost entirely on its own.
Posted
2 days ago
Duration
Format
Demo
hype
Views
20.1K
252 likes
57 · 43
Big Idea
The argument in one line.
DaVinci Resolve's new MCP server lets an AI chat model operate the actual editing software end to end, and the single upgrade that turns a rough cut into a usable one is feeding it a word-accurate transcript instead of Resolve's own.
Who This Is For
Read if. Skip if.
READ IF YOU ARE…
You record long talking-head takes full of false starts and want an AI agent to handle the editing grunt work instead of a human editor.
You already use DaVinci Resolve Studio and want to know exactly what its new AI MCP integration can and can't do right now.
You're comparing AI video editing tools like ChatCut or Descript and want a real cost and accuracy comparison against DaVinci Resolve's approach.
SKIP IF…
You're on the free version of DaVinci Resolve. This integration is Studio-only.
You need multi-camera or narrative editing. This is a single-shot talking-head demo, not a stress test of complex footage.
TL;DR
The full version, fast.
DaVinci Resolve Studio 21.1 shipped an MCP server that lets ChatGPT or Claude drive the actual editing timeline. Justin Brown handed it a 24 minute raw talking-head recording with multiple takes and false starts, gave it a plain-English brief on pacing and zooms, and let it run. The first pass, built on Resolve's own transcription, missed several double-ups because that transcript quietly skips words. Swapping in a word-level Deepgram transcript fixed the timing completely with no other changes. From there the AI added animated titles, then a second on-brand version after reading his sales page, and exported a verified 4K file, all in under 20 minutes of total AI run time. Estimated cost for the whole run: $13 to $19 in standard usage, already covered by his existing plan.
Free for members
Chat with this breakdown — free.
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.
Justin Brown claims DaVinci Resolve's new MCP integration beats every other AI editing tool he's tested, and flags one flaw he'll show how to fix.
00:50 – 02:01
02 · Set Up DaVinci Resolve AI Editing
Requires DaVinci Resolve Studio 21.1. File > Setup AI Assistance connects an existing Claude, Codex, or ChatGPT session with one click.
02:01 – 04:11
03 · AI Video Editing Prompt Setup
He briefs the AI like a human editor: use the last take, keep pacing quick but breathe between sections, add eye-aligned punch-ins and slow zooms.
04:11 – 05:47
04 · DaVinci Resolve AI Base Edit Results
First pass cuts 24 minutes down to 7:14 in about 15 minutes, correctly framed and zoomed, but a closer look reveals missed double-ups and false starts.
05:47 – 07:09
05 · Fix DaVinci Resolve Transcript Mistakes
Resolve's built-in transcription tool silently skips over repeated words, which is why the AI edit missed those mistakes in the first place.
07:09 – 08:28
06 · Deepgram Transcript for Accurate AI Editing
Switching to a word-level Deepgram transcript fixes the missed cuts. The AI stops and asks for explicit permission before touching stored API credentials.
08:28 – 09:40
07 · AI Video Editor Timing Accuracy
With the accurate transcript, every cut and pause lands correctly, no further correction needed. Transcript quality, not prompt complexity, was the deciding factor.
09:40 – 11:23
08 · Add DaVinci Resolve Effects and Graphics
Given zero style guidance, the AI generates animated title cards timed to the transcript's key points in under 9 minutes.
11:23 – 13:05
09 · Match AI Graphics to Brand
Pointed at the creator's own sales page, the AI reads its fonts, colors, and layout, then rebuilds the whole graphics package to match as a separate duplicate timeline.
13:05 – 14:02
10 · Export 4K Video from DaVinci Resolve
The AI handles export settings unattended, renders a 4K file to the desktop, and verifies resolution, frame rate, and audio before confirming done.
14:02 – 16:15
11 · DaVinci Resolve Studio Requirements
Requires the Studio (paid) tier and a genuinely capable local machine. ChatCut is named as a comparable alternative workflow.
16:15 – 17:46
12 · AI Video Editing Cost Breakdown
ChatGPT estimates the full run at $13 to $19 in standard usage (already covered by the existing plan) or $30 to $40 at fast-mode rates.
Atomic Insights
Lines worth screenshotting.
DaVinci Resolve Studio 21.1's new MCP server lets ChatGPT or Claude drive the actual timeline: cutting, framing, titling, and exporting, not just suggesting edits.
Resolve's built-in transcription tool silently skips words during false starts and retakes, which breaks any AI edit that depends on transcript accuracy.
Swapping in a word-level Deepgram transcript fixed every timing and double-up problem the first AI pass missed, with no other prompt changes.
A 24-minute raw talking-head recording with multiple takes was cut to just over 7 minutes by the AI in about 15 minutes of processing.
The AI added eye-aligned punch-ins and slow zooms on its own, keeping the subject's eyes in a consistent position across cuts without being given exact coordinates.
Given a link to the creator's own sales page, the AI scanned it for fonts, colors, and layout, then rebuilt the graphics package to match the brand in under 10 minutes.
The AI refused to retrieve or use stored API credentials on its own. It stopped and asked the human to authorize Deepgram access instead.
The tool verified its own export (resolution, frame rate, audio channels, file size) before reporting the job complete.
Estimated cost for a full run, transcription through export, came to $13 to $19 in standard API-equivalent pricing, or $30 to $40 at fast-mode rates.
The creator briefed the AI the way he'd brief a human editor: which take to prefer, how tight to cut, and where to leave breathing room between sections.
The one named flaw, an inaccurate built-in transcript, is described as the same reason other people are already reporting bad results with this same integration.
Takeaway
Feed the AI editor an accurate transcript first
AI EDITING WORKFLOW
An AI agent can now run a real video editor end to end, but the entire result hinges on the accuracy of the transcript you feed it, not the sophistication of the prompt.
02Set Up DaVinci Resolve AI Editing
The feature requires DaVinci Resolve Studio (the paid tier) and version 21.1 or later. The free version does not have it.
Setup is a single menu action that connects your existing Claude, Codex, or ChatGPT session to the app. No separate account or plugin install is needed.
03AI Video Editing Prompt Setup
Brief an AI editor the same way you'd brief a human one: which take to prefer, how tight to cut, and where pacing can loosen between sections.
Telling the model your recording habits, like false starts and re-attempts spoken off the top of your head rather than word-for-word, helped it interpret ambiguous footage correctly.
A single basic prompt, with no follow-up refinement, was enough to get a usable first cut in about 15 minutes.
04DaVinci Resolve AI Base Edit Results
The AI cut 24 minutes of raw footage down to about 7 minutes on the first pass, correctly identifying which takes to keep.
It added eye-aligned zooms and jump cuts on its own, without being given exact framing instructions, keeping the subject's eyes in a consistent screen position.
A first pass that looks polished on a quick scrub can still hide small errors. Only a closer watch caught them.
05Fix DaVinci Resolve Transcript Mistakes
DaVinci Resolve's built-in transcription tool silently drops words during false starts, so edits built on it inherit that same blind spot.
The AI couldn't find or fix the missed double-ups on its own, even when asked directly, because the transcript it was working from was already wrong.
06Deepgram Transcript for Accurate AI Editing
Swapping in a word-level Deepgram transcript, with no other prompt changes, fixed every mistake the first pass had missed.
The AI stopped and asked for explicit permission before touching stored API credentials, rather than trying to retrieve or use them itself.
07AI Video Editor Timing Accuracy
Transcript accuracy, not prompt sophistication, was the single biggest factor separating a rough AI cut from a genuinely clean one.
With the better transcript, the AI matched professional pacing on pauses, breath removal, and section breaks without further comment or correction.
08Add DaVinci Resolve Effects and Graphics
Given zero guidance on style, the AI produced animated title cards timed to key phrases in the transcript, understanding context it was never explicitly told.
A graphics pass that would take hours by hand in a template tool took under 9 minutes end to end.
09Match AI Graphics to Brand
Pointing the AI at a live webpage let it extract real fonts, colors, and layout, then rebuild the graphics package to match, with no style guide or asset export.
Asking for a duplicate branded version rather than an overwrite preserved the original edit, so both versions exist side by side to compare.
10Export 4K Video from DaVinci Resolve
The AI handled export settings, like bitrate, resolution, and naming, without being told any specifics, then verified the finished file itself before reporting done.
11DaVinci Resolve Studio Requirements
Running this locally requires a genuinely capable computer, since the editor, the render, and the AI session are all happening on the same machine at once.
This is one of several competing AI editing workflows right now, not a Resolve-exclusive capability.
12AI Video Editing Cost Breakdown
A full run through the workflow, transcription through one branding pass and export, was estimated at $13 to $19 in standard usage, or $30 to $40 at fast-mode rates.
Those costs were already covered by an existing subscription plan rather than billed separately, which changes the real-world math significantly.
More upfront context and guardrails in the prompt produce more consistent, repeatable results across videos shot the same way, at the cost of some flexibility.
Glossary
Terms worth knowing.
MCP server
A protocol that lets an AI chat tool like ChatGPT or Claude send commands directly into another piece of software, in this case DaVinci Resolve's editing timeline, instead of just generating text back to the user.
Deepgram
A paid speech-to-text API that returns word-level timestamps in a full JSON transcript, used here in place of DaVinci Resolve's own transcription because it doesn't skip over false starts.
Punch-in
An edit that zooms into the same shot without cutting to a different camera or angle, used to add visual movement to an otherwise static talking-head recording.
ChatCut
A separate AI-driven video editing tool the creator has used before, named here as an alternative to DaVinci Resolve's new MCP-based workflow.
Resources
Things they pointed at.
00:50toolDaVinci Resolve Studio 21.1
07:09toolDeepgram
14:02toolChatCut
00:00toolGPT-6 Astra Medium (via Codex App Tools)
Quotables
Lines you could clip.
00:08
“This is by far one of the coolest things that I have seen.”
“We don't just want amazing. We want really fucking amazing where we don't need to touch this at all.”
blunt standard for what 'good enough' AI output actually means→ IG reel cold open↗ Tweet quote
07:09
“I think this is the flaw that I saw previously with the DaVinci Resolve transcript tool. It skipped over mistakes.”
names the exact bug other users are hitting→ newsletter pull-quote↗ Tweet quote
15:25
“It's not just doing the work, it's checking its work and making sure that everything is all good.”
captures the self-verification behavior that's the real differentiator→ TikTok hook↗ Tweet quote
The Script
Word for word.
Read-along
Don't just watch it. Burn it in.
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.
17px
analogystory
The latest version of DaVinci Resolve 21 .1 is now out and it is a massive step up. As someone who tests a lot of these different AI tools and functionalities, this is by far one of the coolest things that I have seen. They now have an MCP server.
Meaning that you can use your AI tools like Claude, Codex, ChatGPT to control your timeline. And yes, we've seen a lot of other tools have this kind of functionality in their chat cut is a big one that I'm a fan of. Descript as well has this functionality, but to me having this in a more mainstream video editing tool, a production level tool, I was pretty skeptical about how good it would actually be.
And it's totally blowing my mind. There is one big flaw with it, which I will cover off because this is where I'm seeing so many people have issues online with this already. But I'm going to show you how to fix it as well.
But let's jump across to DaVinci Resolve. So yes, you do need to be on the studio version, the paid version of this to access this. And you do need to be on the latest version, 21 .1.
And we can then come up to file and you come down here to set up AI assistance. Just clicking this automatically goes through. You can see my...
Claude is gonna pop up here and it's gonna automatically connect. So I'm just gonna cancel out of this because I already have it connected, but there's nothing more that you need to do. It's already done at this point.
Maybe close, reopen your Claude or your ChatGPT if you had it running, but then we can just call on it from those AIs. So I've been trying this a lot and with other AI tools as well to work out which AI is the best for being able to edit our videos down, irrespective of the tools like DaVinci Resolve, ChatCut or whatever.
And right now Astra is just totally... blowing everything else out, Fable included, out of the water. So I'm gonna come down to my chat GPT here and make sure we scroll right back to the top.
So this is yesterday, a test from yesterday when I installed this again on this system. And I said, as far as I know, I've just connected the chat GPT MCP for DaVinci Resolve. Can you verify it's working?
I just wanted to make sure it was all good. It says, yes, it's working, awesome. Okay, so I'm giving this super basic prompts and obviously with anything AI, the better the prompt, the better the output, but I always just wanna see, If I give this something pretty basic, what can it do with it?
Okay, so given that we're already talking about DaVinci Resolve here in this chat, I've just said, okay, great. I want you to start a new project and let's edit this file down. So I've linked it to the file or just dropped the path of the file in.
Obviously you can actually go into DaVinci Resolve, do this manually if you'd like, but hey, let's let the AIs do as much as they can. This is a basic talking head video for a sales page for a new video operating system intensive workshop. So a workshop that I'm gonna be running very soon.
So there were multiple attempts at some of the different paragraphs. There's quite a few false starts. I did leave a few notes as to which take to use or notes around this, or I'll do this again as I was recording.
So I'm kind of talking to myself or talking to an AI editor when I'm recording. I don't use this one. Actually, I'll do this again.
Those kinds of things just in line with the recording. So take a look at those, but as a general rule, I want you to use the last take or the last attempt, even if the words that I said have changed. slightly from the previous one.
I wasn't using a teleprompter for this. So if I'm re attempting something or make a mistake, I'll just go off the top of my head and say something similar, but the words aren't always going to be the same. I also want the pacing to be relatively quick, but okay for you to have more breathing space between some of the cuts at the end of a section.
So when it feels like I'm pivoting from one thing to another, the cuts don't need to be as tight. I guess I'm giving it some flexibility in there, but also some guidelines. I'd like you to also add any jump cuts, zoom in, on the shot.
Make sure to keep the person's eyes in the same or very close to the same position. So it's less jarring for the viewers. Also added any slow zooms to make this more interesting and engaging in critical areas of this sales video.
So again, these are things that I would tell a human editor. Here's what I shot. Here was my intention.
I did do some multiple retakes and at the end product, I'd like you to have slow zoom ins and that kind of stuff. Now, normally I wouldn't go that far, but hey, I'm just testing this stuff. And so this was with the ChatGPT -6 Astra Medium.
I didn't even have this on the highest setting. It ran for 15 minutes and 30 seconds. So already much quicker than a lot of the other AIs.
Then it did an awesome base edit, which I'll bring up in just a second. But there is one piece that I did call out while it was working. So you can steer it, or you can kind of ask it questions while it is working.
And I said, while this is going, I can see that it's using the built -in transcribing feature in Resolve. So Resolve does do its own transcribing. In the past, this hasn't been very accurate.
Is it at all possible to use a different service for this? Let's continue with this one for now, but more of a question, is this possible? And it's come back and said, yeah, it is using it and you can use another AI because that to me is the biggest issue, but let's take a look at what it actually made.
So let's open up the first edit here. First off. I haven't touched anything with this timeline or this entire project.
So the timeline that's generated, the file that brought in, all the graphics and things that I'll show you as well that we made further down the process, all totally hands off from me. So this is the edit. Let me zoom out on this.
You can see it's performed all the cuts. The raw footage, if I open that, is 23 minutes 58 versus what this cut down to at just over seven minutes, seven minutes 14. So it's...
done a lot of what I asked here. And already at this point, I was pretty blown away. And you can see all the markers and things that it's used.
Final take selected, eye framing. And even if I just maybe mute this and scrub through it, look at the slow zoom in, right? So at the start, we want to draw people into the video.
That's there. There's a zoom in there. It's zooming in and out.
The eyes are in the right position. So irrespective of the content and where the cuts are and everything right now, this is already blowing my mind. So pretty awesome.
already this could be engaging but the biggest question then is how did it actually go with the edit and i started to look and it had done some good things and it missed a few things so i'm looking at the start of this video so we play this so today is tomorrow and this works now okay so this little bit all right so it's not bad Because it has picked the cut that I want it to use, but I want it to be cut from here.
This works now because I... So you could say, okay, this is amazing. But we don't just want amazing.
We want really fucking amazing where we don't need to touch this at all. But this to me was also a bit of a red flag. I start to look down like, well, what's this little chunk here?
Why would this be in our finished video? So I'm just quick gauging stuff at this point. And let's play this section.
And when I looked at where... And when I looked at where my time, okay, so it's missed something. It missed a double up, a false start, a mistake.
And I do that all the time. I look down here and there's another one. I'm like, okay, so it's got the general gist of it and it's done a pretty good job, but it's missing these double ups.
And this is something that I found with DaVinci Resolve's transcribing tool is that it seems to not list out every single word. that you say, if it's something like this, it will skip over the first bit and just give you the text for the second bit. Meaning that when it's applying its edits and things, it's actually not got the highest level of accuracy that you need to have an accurate cut, which is why I was asking ChatGPT which one it was actually using.
So I come back here and said, okay, that was a good edit. Like the rest of it, the takes that it picked and everything was perfect. So the edit was really good.
I did notice watching it back that there were probably three or four failed attempts at the start of some of the sentences. So I started to say something and then I'd say it again, but maybe it happened quick so it was hard to detect. Also, if you look at the audio waveforms, a couple of cuts, there's a breath or some noise just before it.
We want to trim those out. And when I submitted that, it was pretty much saying that it couldn't find those. And to give me specific...
times for those which to me i thought was pretty weird and then i thought i'd just give it some context and say okay i think i've seen this before with the transcript tool it skipped over mistakes let's use deepgram which is the one that i found is the highest level of accuracy it is something that you do need an api call to i think we've paid 20 in terms of credits to deepgram and i haven't needed to top it up and we've done a lot of content through there so i said let's bring in the transcript then from deepgram so i'd already had the transcript done for that file through Deepgram.
So I gave it the files for that. And it's like crazy. It's not just one little text file.
It's a full JSON file with gaps with each individual word timestamp. So the level of accuracy is huge. So it says here, okay, those files worked.
I used Deepgram and it tidied up a new timeline just from me giving it a more accurate transcript. Let's take a look at this one. All of those little cuts and things are perfect.
The spacing, the timing of everything. didn't even comment on the timing any more than that original post. So let's take a look at some of these.
To what goes out and what doesn't. So I've been doing this for 20 years from running a production company. All right, they're all good.
They're all timed perfect. The pacing is good. And this has really been the hardest piece with getting AI to accurately edit your video consistently is the timing piece.
So that transcript level of accuracy was really the piece that was missing here from just the DaVinci Resolve workflow. So let's take another look. Processes and stuff too.
Now, obviously I'm still guiding the research and bringing my experience, my thoughts and opinions. Right, so this was solid. I'm like, wow, that was amazing.
So really in... the original 15 minutes plus four minutes 47 when we switched to the other transcript, which we could just call it and tell it to do from straight away. I had a full base edit with no changes, no tweaks, no adjustments from me at all.
And it checked its work in that time as well. It rendered something out. So there's a seven minute six video and it passed all those checks.
The audio is synchronized. There's no gaps, no black frames. It's all good.
So at this point, we're essentially under 20 minutes and we've got a base edit, but again, from scratch could be faster. Then I said here, can you add some key titles and graphics animated where you think it would help to make this video a great experience for people watching and easy to understand the key points. Now, for some reason that post didn't go and I had to post it again.
So no guidance again around branding or any of that stuff. I tried to keep this clean, just see what it makes without any other. input or anything.
Now I do have Remotion installed on this. I do have Hyperframes installed, but I just wanted to leave it and said, you go make graphics. Can you do it?
What do they look like? Whatever. Let's go.
So again, it worked for eight minutes, 27. And let's have a look at the next one. So this is automatically creating new timelines each time.
So just call the first one sales page edit. The next one was the deep gram cleanup. And then it's made one here, titles and graphics.
Okay. So let's open this. Let's come back across and look at all these titles and graphics that it made.
Didn't give it any guide. but let's take a look at some of them. So Mike asked me to record a quick video for this page.
All right, solid, animated in. It's definitely talking about what it is. I didn't give it the context.
It knows from understanding the transcript. All right, it's timed to what I'm saying. This is just awesome and so fast.
I'll just mute this so I can scrub through it. But I then talk about the 15 % versus the 85%. You can see then we've got another title that builds out over time here.
More titles to build out over time. These look decent. They're not matching our brand in any way, shape or form, but I didn't ask it to or give it any guidance around that.
But the key points, the key messaging and everything, it's all good. It looks awesome. And the time to do these manually in something like After Effects or buy a video hive template or something like that would be huge.
So we're still on a very small time period here. So the next thing obviously I tested was to come back here to ChatGPT and... I say, okay, those are great.
Can we duplicate the timeline? So I don't want you to replace what you've done. I just want to, you know, let's make a whole new version of this timeline and a new set of titles and graphics that match the theme and the branding of the information page that this video is going to be hosted on.
So we have the webpage here, which is the page here. So we can start to see the types of fonts and everything that are used. We can see the graphics and things that are in here.
All right, so I've given it this page and it opened up an AI to come and check this. But if we have a look at what it created then after running for how long? 9 .58, it finished 9 .50.
So eight more minutes to go and recreate, but to scan that webpage and look for branding elements and things. Then the final here, workshop branding. Let's take a look at this.
I'll just mute this audio. So we again have something that matches the colors, the theme, the fonts, the branding. Like this is awesome.
There's the 15%, the 85. So the other one didn't look bad. without context without guidance but this matches the brand this matches the page that it's going to go on and this to me is just the coolest thing that the time to create something that looks decent that is human -led this is an ai slop that's me that's my thoughts and opinions but i'm not needing to do the grunt work and edit this down remove the bad takes them over i can focus on just presenting and doing the bit that i do so we come through here all the graphics and everything the zoom ins even put a call to action at the end build your own version apply for the beta like all of this stuff is on brand is timed up is perfect and then we can even just say okay this is fantastic save a 4k version of this video on my desktop and here it goes it's going off i'll export the workshop branded version in 4k to your desktop you can see in the background like it's actually controlling here it's rendering already so i didn't need to worry about giving it specifics for bitrate
or anything we can see it's named the file while this rendering happened here and we can see this actually happening as if it was a human controlling this computer it's doing it all by itself But it's also checking its work. It's also verifying everything is how you want it and where things technically should be, which is something that's going to lower the bar for people to be able to use and access all the cool features and things that are in pro -grade software like this without needing to truly master and understand where everything is.
So for me, anything that allows people to get their message out there easier, I'm a big fan of. And I think this is a big step forward. So yes, this is something that needs to run on.
the studio version of DaVinci Resolve. And you will want, because it's running on your computer, a pretty decent computer to be able to do this. Now, there are other AI workflows, and we did a video recently around Chatcut, which is a good alternative for this and has a similar sort of workflow.
But you can start to see the types of results and things that you can get as these AI models evolve with less and less. context. And I say that with a caveat, because if you are giving it more context, if you are giving it more guidance as to the things that you want and building processes around these without them being really hard and fast guardrails that it's going to hinder the result that you're getting, then you can consistently create something repeatedly based on how you're shooting your videos.
And that's the type of thing that we're covering in that video intensive workshop where I'm running through that stuff live, along with the other processes and things. that sit around this. Okay, so we can see now that export is complete.
Let me have a look on the desktop. We now have the file here on the desktop. Awesome.
And if we jump back to our chat GPT, the export's done. And it's verifying the file, making sure it's all correct when it's done. This is the coolest thing.
It's not just doing the work, it's checking its work and making sure that everything is all good. And here we go. It's verified.
It's 4K, 30 frames per second, stereo audio, seven minutes, six second file, 3 .2 gig. So to me, watching this space evolve and how fast it's evolving is one of the coolest things because it's allowing people to get the outcome, get their message out there easier without them needing to become an abs... master of tools like DaVinci Resolve.
We can just talk about the thing that we want. We can do the piece that only us as humans can do, share our thoughts and opinions, put our message out there into the world. But we don't now need to be master content creators across all these different tools if we don't want to be.
And you can see that even with just some basic prompts and stuff here, we were able to get something that looked decent in a very short period of time. And that's much faster than the three or four hours that it was. was probably not that long ago in terms of how good these AIs are now.
Now, one last thing you're probably wondering is what did this actually cost? Again, I'm running Astra medium, so not the highest level. And I've said, give me the estimated breakdown on what this would have cost if we went from start to finish once with Deepgram only, because we did do the first pass edit with...
DaVinci Resolves transcribing, which was inaccurate. So if we only did the Deepgram one and we did one set of titles, we didn't recreate that second set that was on brand. So we just created the on -brand ones once.
Right through to the export, what would the breakdown be? It's got the breakdown here, including transcribing in Deepgram, 10 to 15 cents. The reviewing, all of that stuff through to 13 to $19 US.
Likely involve eight to 12 million tokens processed. And it says that these are the standard API equivalent costs. So we're not using the API.
We're using what's included in our plan. So it's not like that was in addition to or anything like that, but it does say that we can also use fast mode rates. If we were doing that, we're budgeting roughly 30 to 40.
So it is a bit of a step up, but I didn't need fast. That was insanely fast already. So that's the kind of stuff I'm covering off in that workshop intensive.
And we'll be running others like that as well in the future. There'll be links in the description box below if you want to find out more. And to keep up to date as we're running more of these things.
in the future as well. Make sure you jump on our email list. There is a link to that in the description.
But if you wanna find out more about DaVinci Resolve or even the other AI tools and things that I am using to run a video led business, check out the description box below and I'll see you in the next video. Cheers.
The Hook
The bait, then the rug-pull.
Justin Brown opens with a claim worth testing: DaVinci Resolve's new AI assistant integration might be the best of its kind he's tried, ChatCut and Descript included. Then he shows the actual chat log and the actual timeline, take by take, to back it up.
Frameworks
Named ideas worth stealing.
16:15list
AI Video Edit Cost Breakdown (5 stages)
Deepgram transcription of the 24-minute source
Review takes, select final attempts, and build the edit
Refine pacing, breaths, sentence boundaries, punch-ins and slow zooms
Inspect the page's branding and create one graphics treatment
Check the finished video and export in 4K
ChatGPT itself estimated line-item costs for a single full run through the whole workflow, in standard API-equivalent pricing.
Steal forpricing out whether an AI-editing workflow is worth setting up for a given video volume
CTA Breakdown
How they asked for the click.
VERBAL ASK
16:30product
“There'll be links in the description box below if you want to find out more... Make sure you jump on our email list.”
Soft pitch for a paid 'Video Operating System Intensive' workshop and an email list, delivered as a natural close right after the cost breakdown rather than a hard sell.
Add Modern Creator as a preferred source and Google shows you more of our breakdowns in Search, Top Stories, and AI Overviews. It only changes what you see, and you can undo it in your Google settings anytime.
Add to Preferred SourcesOpens your Google source preferences with us pre-loaded. Tick the box and you're done.
A full walkthrough of the 2026 YouTube mobile app Shorts studio, from recording and AI-generated clips to captions, stickers, and publishing, no separate editor required.
A full walkthrough of Instagram's free Edits app — a 4K60 camera with teleprompter and green screen, plus an AI editor with cutout and restyle filters.
An editor reverse-engineers 900 of Alex Hormozi's uploads into a rule file, then points the resulting Claude Code system at three unrelated podcasts to see where the rules hold and where they break.
00:33
00:46
01:04
01:09
01:26
01:39
02:06
02:19
02:46
02:59
03:13
03:26
03:39
03:53
04:06
04:19
04:33
04:59
05:13
05:26
05:39
05:53
06:06
06:19
06:33
06:46
07:13
07:39
07:53
08:06
08:19
08:46
08:59
09:12
09:26
09:39
09:52
10:06
10:19
10:46
10:59
11:12
11:26
11:44
11:49
12:06
12:19
12:32
12:59
13:26
13:39
13:52
14:06
14:32
14:46
15:03
15:07
15:27
15:39
15:52
16:06
16:32
16:46
16:59
17:18
17:26
17:39
14:07
15:33
Video of the Day15:47
08:59
11:32
19:21