Everything you need to run this is on this page. Four steps gets your first video into Astra's hands.
Go to ChatGPT → Plugins → search "Remotion" → Install Plugin.
Remotion is the tool Astra uses to build the edit, add graphics and captions, and render your video. Install it once and it's there for every video after this.
Choose Work → GPT-6 Astra → High.
If you want Astra to work with files and apps on your computer, use the desktop app with Work locally selected. Make sure the task can access your recording and your reference files. If it can't reach the folder, upload the files directly into the chat.
Drag the video you want edited into the chat. You can record in sections. Leave a pause between attempts and repeat the complete line when you restart. Tell Astra to choose the clean takes and keep the cuts tight but natural.
Paste the starting prompt from the next section. Then copy the full Video Editing Playbook from further down this page into the same message before you send it. The playbook gives Astra your editing standard.
Alongside the prompt, tell Astra where you'll post the video, your target length, and your exact CTA. For a first Reel, vertical 9:16 and under 60 seconds is a useful starting point. A shorter video is fine when it tells the complete story.
Ask for a finished MP4, captions, and the editable Remotion project so you can review the video and keep working on it.
Copy this exactly. Paste it into the chat with your video attached, then paste the playbook right below it in the same message.
i want to edit a video using Astra - to be published for short form video, where you can take over my screen and use remotion or whatever you need to in order to make it happen. adding in the graphics, b-roll, doing the editing and trimming to make it world class. I have a video editing playbook that you can use for context on my editing style and what to do (see attached). attached is the video I'd like edited. if there's any additional questions or context you need me to give please ask.
You can get started with the four steps above. When you're ready to give Astra more creative direction, add these details in the same chat.
Your reference folder is one of the most important parts of this process. It shows Astra what you mean by great editing: the framing, captions, graphics, pacing, and overall feel. Here's how I collect references:
Start with one primary reference. You might say: "Use this for the caption placement, clean graphics, and pacing." If you add more, explain which part of each one matters.
A reference shows the style to model. It is separate from footage you want included in your video.
B-roll is supporting footage shown while you're speaking. Put usable assets in a separate B-Roll folder and include anything that helps show your work:
Give Astra a simple direction: "When I say [phrase], show [filename]." For a how-to, provide the actual steps in order. The upload, the prompt paste, and the finished result should appear when you talk about each one.
When I say "[spoken phrase]", show [filename]. When I say "[spoken phrase]", show [filename]. When I say "[spoken phrase]", show [filename]. Show these in the order I say them. Use the actual screen recording for each step, and show the real result when I talk about the result.
You can start without extra B-roll. Supplying your own relevant assets gives Astra better material to work with.
I don't just accept the first version. Reviewing and directing the changes is part of the process. This is where the video goes from good to yours.
Watch it all the way through with sound, then check it at phone size. Run it against these five questions:
Does the opening grab attention immediately?
Are the captions accurate, readable, and clear of your face?
Do the cuts feel natural between sections?
Does each visual match what you're saying and show your actual work?
Is the closing text and next action correct?
Give your notes in the same chat. Use timestamp or spoken phrase → exact change. Quote replacement text exactly. Name the file you want used. Tell Astra which parts already work. Send the notes together and ask for the next numbered version.
At 0:00, show the title immediately with no fade. Keep the same font and placement. When I say "paste the prompt", show my actual screen recording. Don't show the prompt during the upload step. Remove the blurry button close-up. Keep the rest of that scene. The captions and pacing in the middle section already work. Keep those as they are. Please send version 2.
Check that your changes were made, then watch the latest exported video once more. Download the MP4, the caption file, and the editable project. The preview helps you review. The MP4 is the file you upload.
Keep the approved project and your reference folder for the next video. You reuse the style while giving Astra fresh footage, new B-roll, and a new CTA.
The second video takes half the time of the first. The fifth one takes a fraction of that.
This is the editing standard I'm sharing with you. Copy the entire block and paste it into the chat with your prompt. Add your own brand preferences as you go.
# Video Editing Playbook: Client Edition ## The standard Create a premium video that is clear, human, and easy to watch. The speaker and the message lead. Every cut, caption, graphic, and sound should make the idea easier to understand. ## 1. Space and visual hierarchy - Keep captions, titles, and graphics clear of the speaker's head and face. If an element touches the head, move it. - Keep space between captions and graphics. Every element needs its own room. - Center standalone B-roll and graphic panels within the composition. Keep text inside a document or interface in its natural alignment. - Protect names, facts, faces, and important controls inside screenshots and footage. - Check the layout at phone size and allow for the destination platform's interface around the edges. ## 2. The opening - Establish the visual hook in the first two seconds. Opening title text should be visible immediately, without a fade-in by default. - Use one font and one color across the opening title. Keep the wording short enough to read comfortably. - Match the visual promise to what the video actually delivers. - Position the title so it feels balanced and never crowds the speaker. ## 3. Keep the frame alive Use either of these layouts when appropriate: - Full-frame speaker with purposeful captions, motion, graphics, or B-roll supporting the teaching. - Speaker in a lower panel, with supporting graphics above and captions in their own clear space. Give important teaching beats a visual treatment. A static talking-head shot should not remain the only visual throughout a teaching section. Keep movement purposeful. Avoid constant zooms or decoration that competes with the message. ## 4. Graphics and B-roll - Apply the clarity test: a viewer should immediately understand which spoken idea the visual supports. - Show real things. Use actual products, tools, documents, recordings, results, and brand assets when available. - Keep the look warm and human, even when the topic is technical. - Use one idea per graphic and only the text needed to communicate it. - Remove abstract shapes, fake interfaces, generic technology imagery, and decorative elements that add no meaning. - Keep static graphics to roughly two or three seconds in a fast edit. Hold longer only when reading or understanding the material requires it. Use a deliberate crop or motion when helpful. - Make screenshots and controls legible. A blurry enlargement is not useful emphasis. - Match visuals to the spoken moment. Do not reveal the next step early. - When demonstrating a process, show the actual action, actual prompt, and actual result. An illustrative graphic must not be presented as evidence of something that happened. ## 5. Cuts and pacing - Remove failed takes, unnecessary repetition, excess dead air, and setup chatter. - When the speaker records in sections, choose the cleanest complete take for each intended beat and join the sections naturally. - Preserve first syllables, word endings, natural breaths, and pauses that carry meaning. - Keep cuts tight without making the speaker sound rushed. Do not apply the same silence-removal rule everywhere. - Preserve the intended meaning and order of a demonstration. - Transitions should support the pace. Remove accidental flashes, one-frame inserts, and jarring changes inside nested video clips. ## 6. Captions and text emphasis - Keep captions accurate, readable, and synchronized to speech. Check names, tool names, numbers, and CTA keywords manually. - Use short phrase groups that viewers can absorb. Give text enough time to register. - Captions transcribe speech. Separate emphasis callouts highlight an important idea rather than repeating everything. - Keep emphasis callouts to roughly one to three words. Use a small number of meaningful callouts. Two to four per minute is a starting guide. - Keep fonts and color hierarchy consistent. Bring callouts in with the relevant phrase or shortly after it starts, and remove them when the idea changes. - The opening title is its own treatment and appears immediately. ## 7. Zooms - Punch in on a strong line or insight. Pull out for a transition or energy reset. - Use roughly one zoom every eight to ten seconds as a starting ceiling. Shorter, faster edits may need a different cadence. - Prefer subtle movement: approximately 100% to 115 to 120% maximum for speaker footage, provided image quality holds. - Treat zooms as punctuation. Crop software demonstrations separately for legibility. ## 8. Sound - Keep the original voice clear, natural, and consistent across cuts. - Match optional music to the speaker's energy and keep it underneath the dialogue. - Use sound effects sparingly and deliberately. - Apply cleanup only when it improves the recording without damaging the voice. Avoid clipping and harsh processing. ## 9. Closing and CTA - Make the requested next action obvious and use the supplied wording accurately. - Keep captions low and balanced without crowding the face or CTA. - If the CTA refers to a supplied resource, show the real resource when useful. - If no resource asset is supplied, or the brief calls for text only, use a clean text CTA. Do not invent a guide cover or offer mockup. ## 10. Match the format - Short-form social: immediate hook, active visual support, tight natural pacing, and a clear CTA. - Long-form public video: intentional cuts, emphasis on key moments, and enough room for the story to develop. Three to five purposeful cuts per minute can be a starting guide where the material supports it. - Client teaching modules: simpler presentation, fewer cuts, and occasional emphasis. Understanding takes priority over visual density. ## Final quality check - [ ] The opening works immediately. - [ ] The speaker's message and energy remain intact. - [ ] Captions, titles, faces, and graphics have clear spacing. - [ ] Every visual supports the words being spoken at that moment. - [ ] Important text and controls are readable at viewing size. - [ ] Cuts preserve words and natural delivery. - [ ] No accidental flashes, mismatched inserts, or duplicate audio remain. - [ ] The CTA is accurate and easy to follow. - [ ] The exported video has been reviewed from beginning to end. Polish serves clarity. When an effect gets in the way of the message, simplify it.
These are the rails that keep the process from producing a generic edit. Build them in from your first video.
Astra edits toward what it can see. One clear reference with a sentence about why you like it beats five references with no direction. Pick the one. Say why. Build from there.
The first version is a draft. Your notes, in timestamp or spoken phrase to exact change format, are what turn it into your video. The gap between version one and version two is where the taste lives. Keep that gate for every video.
The B-Roll folder is a power-up. Your first video runs on the prompt, the playbook, and your raw footage. Add references and B-roll once you've seen one clean edit come back.
Ship one video. Then make it sharper.
Undeniable Studio is a live building room where you co-create with AI every Tuesday and leave with something real. You just got the full recording-to-finished-video process. Inside, you build the next one live with me, wired to your business, your voice, your stack.
Join The Studio →