Record your screen once. Then ask Claude, or another AI app that runs MCP servers on your PC, for the video. It writes the narration, and Reelquill AI does the rest on your Windows PC: the voice, the cuts, the captions, a presenter, the Shorts. Even a PDF guide, if you want one.
Real footage of the Reelquill AI editor on a silent screen recording of Acme, a made-up app; its captions come from an AI-voice narration. We added the pointer and its click rings, sped up the wait while it makes the video (about 85×) and added a short fade where it loops. Silent.
AI assistants (MCP)
Presenter from a short video
Your own voice
Studio sound
Re-record a part
Share links
Personal copies
Folder actions
PDF & Word guides
Narration from text
Edit by text
Remove pauses & ums
AI captions
Auto-zoom
Shorts finder
Styles
Title & end card
Backgrounds
Blur private areas
Music ducking
MP4 · WebM · AV1
GIF · MP3
AI assistants (MCP)
Presenter from a short video
Your own voice
Studio sound
Re-record a part
Share links
Personal copies
Folder actions
PDF & Word guides
Narration from text
Edit by text
Remove pauses & ums
AI captions
Auto-zoom
Shorts finder
Styles
Title & end card
Backgrounds
Blur private areas
Music ducking
MP4 · WebM · AV1
GIF · MP3
AI assistants (MCP)
Ask once. Get the video and the Shorts.
Reelquill AI gives your MCP AI assistant 37 actions on your PC. It reads the transcript, writes the narration, picks a voice, does the edit and has every file made. You go back to building the thing the video is about.
There’s no chatbot inside Reelquill AI. You talk to the AI app you already use, like Claude Desktop, Claude Code, Cursor or VS Code, and it works Reelquill AI through MCP (Model Context Protocol), the standard way AI apps use tools on your PC.
Step 1
Record it once.
Click through your product like you always do. Talk, or don’t say a word. Your MCP assistant can start the recording too, but Reelquill AI asks you to allow it first. Every time.
Step 2
Ask your MCP assistant.
In Claude Code, Claude Desktop, Cursor or VS Code. Or pick one of the ready-made requests: Clean up a recording, YouTube package, Shorts from a recording.
Step 3
Get the files.
The edited video and every Short land next to your recording. The original stays exactly as you recorded it.
Your AI assistant (MCP)
MCP · reelquill-ai
You
Make a product video from ‘export tour’. It has no sound, so write a short narration and read it in my voice. Add the presenter, music, a title, an end card with our site, and two Shorts.
list_recordingsexport tour.mp4 · 1:52 · no sound of its own
set_narration14 lines · voice: Me (your voice)
check_jobnarration ready · one line held the picture 1.4 s
find_shorts → set_shorts2 Shorts, each with a title
set_music · set_title · set_end_cardquiet bed · “Export in one click” · example.com
set_presenterHeadshot · a ready-made presenter
create_videostarted
check_jobpresenter: 2800/2800 frames, done
files savedexport tour (edited).mp4 · export tour (Short 1).mp4 · export tour (Short 2).mp4
Assistant
Done. The video and two Shorts are next to your recording. Want a YouTube title, description and chapters to go with them?
Recordingexport tour.mp4 · silent
Narrationyour own voice · captions as written
Editcuts · captions · music · title · end card
PresenterHeadshot · from a short video
Filesthe video + 2 Shorts
edited video
Short 1
Short 2
An example session, sped up a lot. For real, the presenter takes about 30 min per minute of video on a laptop without a graphics card, so this one is about an hour. Start it before lunch.
Narration from text · set_narration
Recorded it in silence? Your MCP assistant writes the narration.
Plenty of good demos are recorded without a word. Your MCP assistant writes the lines (or uses yours) and puts each one at the right moment. An AI voice reads them, a stock one or your own. If a line runs long, the picture waits for it. The captions show the words as written, not as heard.
Each line at the time your MCP assistant sets
Your own cloned voice or 10 built-in voices
Captions spelled as written: “manifest.json”, not “manifest dot jason”
Rather do it by hand? Type AI-voice clips in the Voiceover window.
Real footage of the Voiceover window: the lines an MCP assistant wrote for a silent screen recording of Acme, a made-up app, read by the stock voice Heart. We added the pointer and its click rings, left the microphone name blank and added a short fade where it loops; nothing is sped up. Silent here.
It keeps going while you’re away
Long jobs don’t care if your MCP assistant stops waiting. It picks them up again with check_job, and you can watch or cancel them in the app. Videos go in one queue, made one at a time, and the queue survives a restart. Close the window and Reelquill AI can carry on from the tray by the clock, then tell you with a Windows notification when your video’s ready.
Good to know: Reelquill AI has to be running with AI assistants (MCP) switched on. Allow its actions in your AI app beforehand, or the app will stop and ask. Keep the PC awake, and unlocked while it records: Windows blocks screen capture on a locked PC. Editing carries on either way.
Also in the box
Seven more things. No subscription for any of them.
Each of these is usually another app, or another monthly plan. Here they’re in the box, and each one comes with its small print.
Before
After
Studio sound · Light
New
Studio sound
Recorded next to a loud fan? Switch on Studio sound and most of the hum, keyboard clatter and traffic drops away, plus some echo. Pick Light or Strong, then use Hear before and Hear after to judge it yourself. It takes about 2 seconds per minute of recording on our laptop. The catch: music and alert sounds in the recording go too, so it starts off.
Re-record this part
New
Re-record a part
Your product changed and one screen in the demo is now wrong. Don’t redo the whole video. Select that bit on the Timeline, press Re-record this part and record it again. When you Create, the new part goes in its place with the same look, its own captions, and zooms on its own clicks. Up to 50 parts per recording. One quirk: the preview of the whole video still plays the old part.
your-link#key
ViewsHow far they gotPhone · computer
New
Share links Cloudflare can’t peek into
Share a finished video as a link that plays in any browser. It’s encrypted on your PC, then stored in your own free Cloudflare account. The key sits in the link after the #, which browsers never send, so neither Cloudflare nor we can watch it. My shares shows views and how far people got. You can stop a link at once.
Hi Sam!Hi Dev!Hi Maria!
New
Personal copies
Paste a list of names, or open a CSV of up to 500 people, and one demo becomes a copy for each of them. Each one opens on their own card while a voice says hello by name, and ends on their own end card. Say as fixes names the voice gets wrong. The video is made once; then each copy takes seconds (a minute or two each if the presenter says hello).
Old demos12 videos
Captions
Cut pauses
Studio sound
9:16
Even volume
Music
Convert
New
Folder actions
Got a folder of recordings that all need captions? Pick the folder and tick what to do: captions, cut pauses and ums, Studio sound, vertical 9:16, even volume, music, or convert to MP4, WebM or AV1. Then go make a coffee. Each video gets its own place in the queue, and the new copy is saved next to the original. It does that folder only, not the folders inside it.
Step 4Click Export
PDFWordAnidoc
New
Step-by-step guides from the same recording
Press Guide and every click you recorded becomes a step in a PDF or Word guide, with a picture of the screen, the click marked and the button’s name read off the screen. A 33-click recording turned into 17 pages in 136 seconds on our laptop. Anidoc goes further: the pictures move, as a web page or a Word file. Bring your own Claude or Gemini key if you want the words written for you, or let your MCP AI assistant write them. Our logo sits top left, or yours, or none, and a small “Created with Reelquill AI” seal goes at the end. You can switch it off. Guides need a recording made in Reelquill AI, because they work from its clicks.
New
Themes, a tray icon, and quiet updates
Pick Aurora (the default), Neon or Classic. Reelquill AI can sit in the tray by the clock, where videos keep being made and your shortcuts and MCP AI assistant keep working. Updates download in the background and install when you close the app, never in the middle of a recording or a render. You can switch that off. Honest note: it isn’t code-signed yet, so Windows may say “Unknown publisher” the first time.
Real footage of the editor, switched with its own Theme setting: Aurora, then Neon, then Classic (the quietest), then back to Aurora. Short crossfades added.
The presenter is new too: it’s now made from a short video of a person (yours, or a ready-made one). Here’s how it works.
What it replaces
Ten jobs you’d normally do by hand
Making a product video by hand is a pile of small jobs. Here’s who ends up doing each one.
Each job in making a product video: who does it by hand, and who does it with Reelquill AI and your MCP AI assistant
Task
By hand
With Reelquill AI + your MCP assistant
Click through your product on camera
you take after take
you, once one take, and it can be silent
Write, time and record a voiceover
you at a microphone
assistant writes it; an AI voice or your own cloned voice reads it
Someone on camera
you lights, retakes
presenter from a short video of someone who agreed, or a ready-made one
Find the Shorts and reframe them to 9:16
you scrub and crop
automatic Shorts finder + 9:16 that follows the pointer; the MCP assistant picks and titles them
Cut silences and ums
you on a timeline
automatic Remove pauses and filler words; Studio sound for a noisy room
Captions, with product names spelled right
you type them
automatic on your PC; the MCP assistant fixes product names
Redo the parts you fumbled
you re-record
assistant cuts them between words. Or you re-record just that part, or delete the words in Edit by text
Zoom in on what you click
you keyframes by hand
automatic 6 strengths, from Minimal to Close
Title, music, end card
you one by one
assistant sets them
Export each shape and each Short
you one export at a time
automatic one press of “Create edited video” makes them all
you by hand assistant your MCP AI assistant automatic Reelquill AI you, once one step left to you
Measured, not promised
21 sthe transcript of a 110-second talk
35 sthe edited video (106 s) and a 1080×1920 Short, made
31 sa whole 110-second talk re-voiced in a cloned voice (27 lines)
136 sa 33-click recording turned into a 17-page step-by-step guide
All measured times
All measured times, on the same laptop
What
Time
Captions
a 10-minute video in about 1 minute
Built-in AI voice
1–3 s of work per second of speech
Your own voice
about 3× faster than real time (on mains power)
Zoom on clicks
a 10-minute full-resolution video in 3–5 min
Studio sound
about 2 s per minute of recording
Presenter face
about 30 min per minute of video (every frame made)
A new presenter, prepared once
about 3–4 min (for a 13.6-second video)
Presenter look change
seconds
Personal copies
3 copies plus the video in 40 s (voice only, a 6-second test video)
Join 15 clips
a 50-second video in about 11 s
AV1 export (1080p)
about 2.5× the video’s length
All measured on one ordinary laptop with no graphics card: Intel Core i5-1235U, Iris Xe graphics, 16 GB. We don’t have graphics-card timings yet. Your PC will be different, and the app shows its own estimate before it starts.
For teams and builders
Ship the feature. Your MCP assistant ships the video.
You build with Claude Code, Cursor or VS Code anyway. So when a feature ships, record it once and ask for the demo video, the Shorts, the YouTube description and the how-to guide. It’s all made on your own PCs, with no per-video credits. (Your MCP AI assistant’s own plan still applies.)
What your MCP assistant gets: 37 actions
Read 9
list_recordings
get_recording
get_transcript
list_styles
list_voices
list_voice_clips
list_presenter_photos
list_music
list_parts
Edit 11
set_edit_options
cut_parts
make_captions
edit_captions
set_title
set_end_card
find_shorts
set_shorts
use_style
set_music
replace_part
Voice 2
set_voice
set_narration
Presenter 3
set_presenter
preview_presenter
remove_presenter_photo
Batch and copies 3
folder_actions
caption_folder
set_personal_copies
Guides 2
make_guide
get_guide_steps
Record 2
start_recording
stop_recording
Make, follow, show 5
create_video
check_job
list_queue
show_in_app
remove_part
+ 3 ready-made requests: Clean up a recording · YouTube package (title, description, chapters) · Shorts from a recording
How it connects
MCP: no network port
No server of ours
No account
Off until you switch it on.
Connect Claude Desktop in one click, or copy the settings for Claude Code, Cursor and VS Code.
Your recordings folder, and that’s it.
It works with the videos and music in there. Point it at any other folder and the app asks you first.
It deletes nothing without asking.
Replacing your voiceover, removing a presenter or throwing away re-recorded parts: Reelquill AI asks you (Allow or Deny) first. Same for starting a recording, using your cloned voice or adding a new presenter.
Its actions never download or upload.
Every AI model is already in the installer. And there’s no share action, so an MCP assistant can’t send a video anywhere.
Consent for every presenter video,
and only voices you made yourself.
You can watch it, and stop it.
Its jobs show up in the app’s queue, with Cancel.
Set up in a minute
1
Switch on AI assistants (MCP) in Reelquill AI.
2
Connect to Claude, then quit and reopen Claude Desktop. For Claude Code, the app copies the command:
claude mcp add reelquill-ai …
Cursor, VS Code and other MCP apps: Copy settings.
3
Ask: Cut the pauses in my last recording and make two Shorts with good titles.
It can press record, too
With start_recording and stop_recording your MCP assistant can record a fresh take. You still say yes first: Allow once, or for an hour, up to 30 days. It never stops a recording you started yourself.
For teams · Automation (MCP)
Reelquill AI for teams
Your AI assistant runs the edit over MCP: 37 actions
Narration in a stock voice or your own cloned voice
A presenter from a short video of someone who agreed
Styles, folder actions and personal copies for whole batches
Made on your own PCs, with no per-video credits
One-time licenses for companies, from one person to a whole team, with every feature and no usage limits: 3 PCs per person, shared by the team.
Write to support@nextgenomni.com and tell us what you’d like to automate. How it compares with the subscription tools: Why Reelquill AI.
Works with apps that run MCP servers on your PC: Claude Desktop, Claude Code, Cursor, VS Code and others. ChatGPT isn’t supported, because it would need an online relay. Reelquill AI has to be running (the tray is fine) with AI assistants switched on.
Presenter New
A presenter made from a short video.
Give it 15 to 30 seconds of someone talking to the camera: you, or someone who said yes. Reelquill AI redraws their lips so they say your narration. Everything else moves the way it did in their video. It looks like a real person because it starts as a video of one: only the lips are made again. No video handy? Pick one of our ready-made presenters, AI-generated people who don’t exist.
The catch: it’s made, not live. On our laptop with no graphics card the face takes about 30 minutes per minute of video. A new presenter’s video is prepared once, which takes a few minutes. After that, moving the bubble or changing its look takes seconds. It goes on the full edited video, not on the Shorts.
0.29 above1 head0.32neck
Only head and neck show
Drawn for this page. In the app it’s a real person’s video, with only the lips made again.
In the edited video
Here’s the new export, in one click.
Framing
Together at most 0.95, so the chin and some neck stay in.
In the app, Try avatar looks… does this on your actual presenter, and Copy hands your MCP assistant the same settings.
15–30 stalking to the camera
What video to give it
Fifteen to thirty seconds of someone talking to the camera, head and shoulders in shot, a little room above the head, even light. MP4, MOV, WebM, MKV and friends, up to 1 GB; only the first 30 seconds are used. If the video is a bit tight for a look, the app fits the look to it and tells you what was missing.
Face≈ 30 min per minute of video
First time≈ 3–4 min, once per presenter
New lookseconds
How long it takes
Every frame gets made, 25 a second, at about 1.2 to 1.3 seconds each on our laptop’s built-in graphics. That’s about half an hour per minute of video, so a two-minute demo is an hour. Not quick, we know. A 6-second preview lets you check the framing before you commit.
Windows 10/11 64-bit
DirectX 12 graphics
8 GB memory
1.5 GB free
Speed test
Ready for the presenter
It checks your PC first
It puts what the presenter needs next to what your PC has, then runs a short speed test. Its models came with the installer. If your PC would take more than about 75 minutes per minute of video, the app tells you plainly and keeps the presenter off, rather than leaving you staring at a progress bar.
Who is this, and did they agree?Me, and I agree
Consent built in
A video only goes in with a note saying who it is and that they agreed, and your MCP assistant can’t skip it. The ready-made presenters don’t need one: they’re AI-generated people, used under a commercial license. Anything made with the presenter is marked as AI-generated in the file, and you can add a visible label too. Where the law asks, say the presenter is made by AI.
The real app
A look inside the app
A minute of the pre-release Windows app at work, then screenshots. Click a screenshot to see it full size.
Watch it record and edit (1 min). Start recording, the countdown and the recording bar, then the editor: a background, a caption style and Create edited video. These are the app’s own screens from the pre-release app; the pointer, the zooms and the window frames are added, Create is sped up, the narrator is an AI voice, and Acme is a made-up app. It shows a few of the features: see them all.
Do it yourself
The same studio, by hand.
Rather drive it yourself? Record, make your choices, press Create edited video. Every tab feeds that one button, and your original is never touched. The clips below are the real app at work: short, silent, and each one says what we added.
AI captions
Captions that light up as you talk
Your PC turns what you say into captions and draws them into the video. A 10-minute video takes about a minute. Spot a wrong word? Fix it, and it saves as you type. You get an .srt file too.
Five styles: Classic, Highlight, Pop, Karaoke and Boxed
At the top, the middle or the bottom
Fix any word; saved as .srt, or import an .srt
Made on your PC · English
Real output of Reelquill AI: Highlight captions on a screen recording of Acme, a made-up app. These captions come from an AI-voice narration. Trimmed, with a short fade where it loops; silent here.
Rewrite with an AI voice
Said it wrong? Just type it right.
Fluffed a line? Type over your transcript and an AI voice reads the new words in place of yours. Use one of 10 built-in voices, or your own. If the new line needs more time, the video waits for it. All on your PC.
10 built-in voices, or your own cloned voice
The video waits where your new line is longer
Or cut words: delete them, keep the picture, or say it again
Real footage of Edit by text on a rendered demo of a made-up app (not a screen recording), spoken by an AI voice with ums added. The stock voice Heart really made the new line. We sped up the typing about 4× and the wait about 7×, and added the pointer, its click rings, a zoom and a short fade where it loops. Silent.
Shorts
It suggests the Shorts, and they follow your pointer
The Shorts finder picks the moments that make sense on their own, up to 30, 60 or 90 seconds, and tells you why: “Opens with a question · Lots of talking · Busy on screen (6 clicks)”. It’s our own scoring, no extra AI model, so nothing to download. Each Short is 9:16, follows your pointer, and can have your camera right below the screen.
Suggestions with reasons and their own titles
Follows the pointer; camera docks below
The full video and the Shorts from one edit
A real 9:16 Short made by Reelquill AI from a rendered demo of a made-up app (not a screen recording), spoken by an AI voice. The pointer and the red click rings are drawn in that demo, not added by the app. This Short is 0:09 to 0:22, a moment we picked ourselves. Trimmed, with a short fade where it loops; silent.
What the Shorts finder said about the same demo
0:0019 sLots of talkingBusy on screen (5 clicks)Clean start and endBest
Auto-zoom
Zooms in where you click
Reelquill AI zooms in where you click, follows the pointer while you work, and eases back out after a couple of calm seconds. Nobody has to squint at a tiny button.
Six strengths, from Minimal (1.15×) to Close (2.2×)
Follows the pointer smoothly
Add your own zooms anywhere too
Real auto-zoom from Reelquill AI, at the Normal strength. Its source is a rendered demo of a made-up app (not a screen recording), spoken by an AI voice; the pointer and the red click rings are drawn in that demo, not added by the app. Trimmed, with a short fade where it loops; silent.
Look
Put it on a nicer background
Keep the video as it is, or put it on one of 14 backgrounds (gradients, plain colors or a blur of the video itself) with padding, rounded corners and a soft shadow. Crop it first if you like.
Drag the slider to compare
14 backgrounds
Crop freely or to 16:9, 9:16, 1:1, 4:3
Original videoWith background
Two real frames of the same moment: the original screen recording and Reelquill AI’s output on a background. Acme is a made-up app. Only scaled; the original is a little taller, so it has dark bars at the sides.
Also make
One recording →toevery format
Tick “Also make” and one edit gives you every shape you need: 16:9, 9:16, 1:1 and 4:5, plus 3:4, 4:3 and 21:9. And the Shorts.
One edit
16:9 YouTube video
9:16 Short
1:1 Square
4:5 Portrait
Everything else
Everything else that’s in the app
From pressing record to the last export, it’s one app. Here’s the rest of what’s in it.
Record
ScreenWindowArea
REC
Record
Screen, a window or an area
The whole screen, one window or an area you draw (free, 16:9, 9:16, 1:1, 4:3). It records at full resolution, so text stays sharp, and a graphics-chip engine keeps it at a smooth 30 or 60 fps.
Blur or a picture
Record
Webcam bubble
Drag it anywhere, even mid-recording. Blur the room behind you, or put a picture there. The room never flashes up while it starts.
PC
Mic
Record
Computer sound and microphone
What your PC plays and what you say, together. If Windows refuses the computer sound, it keeps recording and tells you.
Alt+Shift+P
RecordingPaused
Record
Pause, resume, 3-2-1 and shortcuts
Global shortcuts start, pause, stop and grab screenshots. The control bar and countdown never appear in the video.
You see
Recording
Record
Speaker notes
Your talking points on screen, scrolling if you like. Only you see them: never in the video, never in a screenshot.
CtrlS
Record
Draw, click ripples and keys
Pen, highlighter, arrows and boxes while you record. A ripple on every click. And the shortcuts you press, if you want them shown; plain typing never is, so passwords stay out.
Edit
0.5×
2×
8×
Edit
Trim, split and speed
Trim the ends, cut out the middle, or split it into separate videos. Slow a part down or speed it up, and fast-forward boring bits at up to 8×.
Soum, todayuh welaunch.
Edit
Remove pauses and filler words
Pauses over 0.5, 1 or 2 s come out, and so do the ums and uhs, even ones the speech model didn’t write down. Check it once on your own voice.
Now open the the Settings menu.
Cut it outKeep the pictureSay it again
Edit
Edit by text
Delete words and the video is cut to match. Or keep the picture and drop just the words. Or say it again: record it, use a file, or have an AI voice read it.
Edit
Zoom on clicks
Zooms in where you click, follows the pointer, eases out. Six strengths, and you can add your own zooms anywhere.
Emailalex@example.com
PlanPro · monthly
Edit
Annotate
Blur or pixelate anything private, and it stays put through every cut and zoom. Add text, arrows, boxes and spotlights.
Edit
Background, frame and crop
14 backgrounds, padding, rounded corners and a shadow. Or crop to any shape.
MP4WebMMOVMKV
Editing
Edit
Open any video later
MP4, WebM, MOV or MKV, from the list, Open a video… or drag and drop. Your earlier edits come back with it.
Voice & captions
CaptionsmadeonyourPC.
ClassicHighlightPopKaraokeBoxed
Captions
AI captions
Made on your PC in 5 animated styles, at the top, middle or bottom. Fix any word; save or import an .srt. English only, for now.
HeartMichaelEmmaFenrir
Voice
AI voices
10 built-in voices, US and UK English, read whatever you type. Use them in Rewrite with an AI voice or Say it again.
Me (your voice)
Voice
Your own voice
Talk for 5 to 20 seconds and it makes a copy of your voice. You tick “This is my own voice” first. Up to 20 voices, English, and it never leaves your PC.
Here is the new dashboard
Voice
Voiceover studio
Talk over playback and redo just the bit you fluffed. Add typed or microphone clips, and let a clip pause the video while it plays.
Voice
Music
Sound
Music and even volume
25 tracks in 8 moods come with the app, or use your own. It gets quieter while you talk, fades in and out and loops without a gap. Even out the volume to −16 LUFS, the level YouTube and podcasts use.
Finish & export
0:12Opens with a question
1:04Busy on screen (6 clicks)
2:31Lively voice · Clean start and end
Finish
Shorts finder
Suggests the moments that work on their own, says why, and lets you title them. Our own scoring, so no extra AI model to download.
YouTube tutorialVertical Short
Video 2 of 5
Finish
Styles and batch
Save a whole look and sound as a style (up to 24, and two come ready-made). “Use on several…” makes an edited video of each recording you pick, one after another. No cap, no credits.
3 things nobody tells youexample.com
Finish
Title and end card
A hook line at the start, and a closing line over a held, darkened last frame.
Finish
Clips and pictures
Put in up to 7 clips or pictures at any moment.
Finish
Join videos and photos
Up to 15 videos and photos with 8 transitions; photos get a slow zoom.
MP4 (H.264)WebM (VP9)AV1GIFMP3.srtA part
Saved as a new file
Export
Export any way you need
MP4 by default, WebM for websites, AV1 when you want the smallest file (about a quarter smaller, but slower). High, Standard or Small quality, full size down to 480p, 30 or 24 fps. Always a new file.
How it stacks up
Cheaper than the usual stack, and slower than the cloud
Most people stitch this together from three subscriptions. Here’s the short version of how that compares, including where we lose.
Reelquill AI next to the usual subscription tools, prices as of October 2026
The usual tools
Reelquill AI
What it costs
A three-tool stack (Camtasia, ElevenLabs and HeyGen), billed every year. See the sums
One price, paid once, with every 1.x update. See pricing
Limits
Credits and minutes on the AI tools. (Descript, for one: its Hobbyist plan gives 10 media hours and 400 AI credits a month)
No credits, no minute caps, no watermark. Your disk is the limit
Where your video goes
Camtasia keeps it on your PC; the AI voice and the presenter go through ElevenLabs’ and HeyGen’s cloud
Stays on your PC, unless you share it to your own Cloudflare, encrypted
MCP AI assistant
Some have an MCP that runs in their cloud and uses your credits
37 actions on your PC, on your MCP AI assistant’s own plan
Where they win
Mac and the web, many languages, team comments, and avatars rendered on their servers instead of your laptop
Windows only, English only, and the presenter takes about 30 min per minute of video
Prices from each vendor’s own pages in October 2026; they change. Sources, and many more apps, on Why Reelquill AI.
Recording, editing, captions, the AI voices, your cloned voice, Studio sound, the guide’s screen reading and the presenter all run on your Windows PC. Nothing is uploaded unless you choose to share a video, and then it goes to your own Cloudflare account, encrypted. With an MCP AI assistant switched on, it gets text (file names, transcripts, captions), never the picture or the sound.
Nothing uploaded unless you share
Your videos, voices and presenters stay on your disk. With an MCP AI assistant on, the text it reads goes to its company, like anything you type to it.
Shared, but still yours
A share link’s video is encrypted on your PC and kept in your own free Cloudflare account. The key is only in the link, so neither Cloudflare nor we can watch it. We run no server for it, so your links don’t depend on us.
No credits, no account
Nothing to buy, count or run out of. No account with us, and no tracking.
Your files stay yours
“Create edited video” always makes a new copy. The original is never changed.
Everything in one download
Every AI model comes inside the installer, so there’s no AI model to fetch while you work. It goes online only to check your license now and then, for updates, and for the things you switch on: share links, an MCP AI assistant, AI-written guides.
.mp4.srtyour voicepresenter
Your PC
Your MCP AI assistanttext only · when you switch it onUploaded only if you share
What you need
No graphics card needed. It warns you before it’s slow.
The app
Windows 10 or 11, 64-bit
Smoothest capture with a graphics chip that encodes H.264 (Intel Quick Sync, NVIDIA NVENC or AMD AMF). Without one, the Standard engine can drop to about 10 fps on high-DPI screens
Captions, AI voices and your own voice: English
Not code-signed yet, so Windows may say “Unknown publisher”
One download, then quiet updates
One installer with every AI model inside
Captions, AI voices, your own voice, Studio sound, guide screen reading and the presenter: included
No model downloads while you work
1.x updates (about 300 MB, the app only) download in the background and install when you close it
The presenter
DirectX 12 graphics: a built-in Intel Iris Xe or AMD Radeon can be enough, and the speed test decides
8 GB memory (16 GB is better) and about 1.5 GB free
Its models come with the install (about 1.0 GB of it)
It records your screen and webcam, then turns the recording into finished videos, Shorts and step-by-step guides on your Windows PC. Captions, cuts, zooms, voices, Studio sound, a presenter: it’s all in there. You can do it by hand, or ask an AI assistant to do it for you over MCP.
Is it available yet, and what does it cost?
Not yet. Reelquill AI isn’t on sale, and the Windows download comes at launch. You’ll pay once, no subscription, for yourself or for a company: see Pricing, and Why Reelquill AI for how that compares with the subscription tools. Want early access, or to talk about automation for your team? Email support@nextgenomni.com.
Where’s the AI assistant? I don’t see a chatbot in the app.
That’s on purpose: Reelquill AI has no chatbot of its own. The assistant is the AI app you already use, like Claude Desktop, Claude Code, Cursor or VS Code. Switch on AI assistants (MCP) in Reelquill AI, connect your AI app once, and from then on you ask in that app (“make the demo video from today’s recording, with captions and two Shorts”). It uses Reelquill AI through MCP, the Model Context Protocol. Don’t want an assistant? Every feature also works by hand in the app.
Which MCP AI assistants work with it?
Claude Desktop (one click on Connect to Claude, then restart Claude Desktop), Claude Code (the app copies the command for you), and Cursor, VS Code and other apps that run MCP servers on your PC (copy the settings). ChatGPT isn’t supported, because it would need an online relay. AI assistants stay off until you switch them on.
Can it really make a video with nobody at the PC?
Mostly, yes. Your MCP assistant can narrate a silent recording, cut it, caption it, add music, a title, an end card and the presenter, find the Shorts and make every file. Long jobs keep going on their own. It can even start a recording now, but Reelquill AI asks you to allow that first, so someone does need to say yes (once, or for a while: from an hour up to 30 days). Allow Reelquill AI’s actions in your AI app beforehand, keep the app running and the PC awake, and keep it unlocked while it records.
Does my assistant see my videos?
Not through Reelquill AI. Its actions give the MCP assistant file names, transcripts and captions, never the picture or the sound, and that text goes to the assistant’s company. One thing to know: an assistant that can open files on your PC by itself (like Claude Code) could open a video if you let it.
How do share links work, and who can see my video?
You set it up once with a free Cloudflare account and a token, which takes about a minute. When you share, the video is encrypted on your PC and stored in your account. The key is the part of the link after the #, and browsers never send that part anywhere, so Cloudflare can’t watch it and neither can we. Links last 7, 30 or 90 days, or until you stop them. My shares shows views, how far people got and which parts were played most, with no cookies and no IP addresses stored. Limits: up to 900 MB per video without a card on file, and we haven’t tried Mac Safari or strict company networks yet.
Is the presenter live, and will my PC run it?
It isn’t live. You give it a short video of someone talking to the camera (or pick a ready-made presenter), and it redraws their lips to say your narration. It goes in a bubble on the edited video, not on the Shorts. On a laptop with no graphics card the face takes about 30 minutes per minute of video. A new presenter is prepared once, in a few minutes, and after that look changes take seconds.
The app checks your PC first: Windows 10/11 64-bit, DirectX 12 graphics (a built-in Intel Iris Xe or AMD Radeon can be enough), 8 GB of memory and about 1.5 GB free. Then it runs a short speed test. If it would take more than about 75 minutes per minute of video, it tells you and keeps the presenter off.
Whose video can I use for the presenter?
Your own, or one of someone who agreed. The app only adds a video with a note saying who it is and that they agreed. Or skip that and use one of the ready-made presenters, who are AI-generated people that don’t exist. Anything made with the presenter, an AI voice or your cloned voice is marked as AI-generated in the file’s metadata, and you can add a visible “AI-generated” label as well. Where the law asks for it (in the EU, for example), say that the presenter or voice is made by AI.
Can it make how-to guides as well as videos?
Yes. Press Guide on a recording made in Reelquill AI and you get a PDF or Word guide with one step per click, each with a picture of the screen and the click marked. The button names are read off the screen on your PC. Anidoc makes the same guide with moving pictures, as a web page or a Word file (a PDF can’t move, so it keeps stills). For the words, bring your own Claude or Gemini key, or ask your MCP AI assistant to write them on your own subscription. Our logo sits top left unless you pick yours or none, and a small “Created with Reelquill AI” seal goes at the end. You can switch it off.
Is it really offline?
The work is. Recording, editing, captions, the AI voices, your own voice, Studio sound and the presenter all run on your PC, and every AI model comes inside the installer. The app goes online to activate your license once on each PC and check it about every two weeks (it never sends your files, and it keeps working as long as one check succeeds in 90 days). It also goes online for updates, and for the things you choose to switch on: share links, an MCP AI assistant, AI-written guides.
Files you export carry a one-way code made from your license key: never the key itself, nothing about you or your PC, and nothing is sent.
Does it update itself?
Yes, that’s how it’s built: every 1.x update downloads in the background (about 300 MB, just the app; the AI models stay) and installs when you close the app, or straight away with Restart now. It won’t restart while something is recording or being made. You can switch updates off, or press Check now.
Which languages?
English, for the captions, the built-in voices and your own voice. The app itself is in English too.
What AI does it use?
Open models that run on your PC: whisper.cpp for captions, Kokoro for the 10 voices, Kyutai’s Pocket TTS for your own voice, DeepFilterNet3 for Studio sound, PaddleOCR for reading the screen in guides, and MuseTalk for the presenter’s lips. All permissive licenses, and their notices come with the app.
Your next video, without the re-shoot
Reelquill AI is coming to Windows 10 and 11. Let your MCP AI assistant do the edit (narration, presenter, captions, Shorts, the guide), or do it all yourself. Either way it’s one app, on your PC, paid for once.
