How to Turn a Long Video into Multiple YouTube Shorts with ChatGPT

One long interview can give you several YouTube Shorts. Turning it into a batch means finding the moments worth using, cutting each story so it makes sense on its own, and repeating the framing, captions and sound work for every video. Sometimes the setup and payoff are several minutes apart, so selecting a promising quote is only the beginning. By the time you’ve finished one Short, you still have to do it all again for the next.

ChatGPT can take on much of that work, but it needs a compatible video editor connected through MCP to carry out this timeline workflow. The connection gives it access to the editor’s controls, so your instructions become changes to the edit. You can choose any editor with a compatible MCP connection; setup and available capabilities will differ.

One free option is invideo Editor. Its MCP connection lets ChatGPT search footage, build separate cuts, add supporting visuals and finish each Short on a professional-grade, frame-accurate timeline. Its full manual editing suite stays available throughout. Ordinary editing through MCP uses your ChatGPT allowance without consuming invideo credits. Generating new images, video or audio uses invideo credits separately.

For this walkthrough, imagine a 38-minute interview with the owner of a neighbourhood bakery, filmed in one wide shot with both speakers visible. You also have footage of pastries being prepared, the shop counter and a handwritten menu. The goal is three Shorts for viewers interested in the stories behind small businesses. The example prompts show how to develop that batch; replace the selections and timings with those your recording supports.

1. Connect ChatGPT to the editor

Open invideo’s connection page, choose ChatGPT and follow the instructions for your account. On the web, open ChatGPT Plugins, select + and add a custom MCP server named “invideo” with this URL:

https://mcp.invideo.io/mcp

Sign in to invideo when prompted, install the connection and select it in your editing conversation. Follow the connection page if your account requires developer mode or workspace permission.

The ChatGPT tab on invideo’s connection page shows the MCP server URL and setup route.

Keep invideo Editor open alongside ChatGPT so you can play the timeline and make manual adjustments as the edit develops.

2. Add your recording and explain the Shorts you want

Upload the interview and supporting recordings to a “Bakery Interview” project in invideo Editor, or attach the files in ChatGPT and ask it to add them through the MCP connection. For an existing project, give ChatGPT its link or name and identify the full interview timeline.

Explain the audience and the kind of batch you want before asking it to make cuts:

The illustration shows the attached recordings and a brief in ChatGPT that establishes the audience and intended batch before assembly.

You do not need to watch the recording and supply every timestamp yourself. ChatGPT can search the footage by spoken content or visual action. Give it enough direction to distinguish a useful passage from an isolated, attention-grabbing sentence.

3. Find distinct stories worth turning into Shorts

Ask ChatGPT to compare candidates before assembly when you are still deciding what the Shorts should cover. That search is a separate job from building the cuts: you are choosing three different ideas from a much longer conversation.

Find five candidates for three distinct Shorts. Look for a decision about pricing, a product launch that did not go as planned, and a customer request that changed what the bakery offered. Include other strong stories if the recording supports them. For each candidate, give the source timestamps, a possible opening, the payoff and any setup or qualification that must stay. Avoid candidates that repeat the same takeaway. Return the shortlist without changing the timelines.

The illustration shows three source sections marked on the full interview timeline. The marker ranges are examples; use the timings from your recording.

Once ChatGPT is done working, watch the suggested source sections on your timeline. A launch story needs the expectation, what happened and what the owner changed afterwards. A reason behind her pricing may need her explanation of the labour involved, rather than just the final price decision. Keep a qualification that changes the meaning of the answer, even if a shorter excerpt sounds more dramatic.

Suppose the footage supports these three selections:

• “The Price Decision”: why the owner changed a pastry’s price and what she included in that decision.

• “The Launch That Flopped”: what she expected from a new product, what actually happened and how she responded.

• “The Customer Request”: a customer’s suggestion, the change she tried and the result she describes.

These are three different editing jobs: a concise explanation, a story with a reversal and a story about listening to a customer. Use the actual interview to decide which passages deliver those outcomes. If you already know the passages you want, you can go directly to assembly with their source references.

4. Build separate cuts and develop their hooks

ChatGPT can create separate cutdowns from those selections. Name the timelines so later feedback applies to the right Short.

Create three separate timelines named “The Price Decision,” “The Launch That Flopped” and “The Customer Request” from the selected passages. Leave the full interview unchanged. Keep the pricing explanation and its qualification, the launch’s expectation and outcome, and the customer request followed by what changed. Remove false starts, repeated points and unnecessary dead pauses. Aim for roughly 40–60 seconds each where the material supports it; preserve a complete idea over a rigid duration. Leave supporting footage, captions and music for the next pass.

The illustration shows “The Price Decision,” “The Launch That Flopped” and “The Customer Request” as separate editable timelines, with the full interview still available.

The first assemblies land on the editable timeline with individual clips and cut points. You can trim a pause frame by frame, slip a clip to choose a different source range, or reorder a passage yourself. ChatGPT reads the updated timeline when you continue.

Now ask it to develop the openings around what each Short delivers:

Suggest two hook alternatives for each Short. Give exact source quotes and timestamps for any interview-audio opening. For “The Price Decision,” compare a direct answer with a brief text question. For “The Launch That Flopped,” compare the owner’s expectation with her reaction to the outcome. For “The Customer Request,” consider opening with the request itself. Do not invent dialogue or promise an outcome the Short does not explain.

You can then apply the choices in one instruction:

Use the direct-answer opening for “The Price Decision,” the expectation-led opening for “The Launch That Flopped,” and the request-led opening for “The Customer Request.” Remove duplicated setup after each hook. Keep the owner’s original meaning, qualifications and complete endings.

Compare the first few seconds with what follows. A question needs an answer. A surprising quote needs enough context to mean the same thing as it did in the interview. If you want to compare two edited openings, ask ChatGPT to duplicate that Short onto another timeline first.

5. Refine a Short with targeted feedback

ChatGPT can target a named timeline and time range while preserving the parts you have approved. For example, suppose the launch cut repeats the owner’s initial expectation after you have already used it as the hook:

The illustration shows feedback targeting the repeated expectation at 00:09–00:15 while preserving the other two Shorts.

Those timestamps are illustrative. Use timings from the current cut, not the full interview, when you want to change the Short’s sequence.

Play the revised join and ending. You can keep a pause because the owner’s expression helps the story land, or trim it manually to suit your pacing. Manual editing also extends to effects, colour, animation and sound. Explain why an approved moment matters when that intention is not apparent from the timeline.

6. Add supporting footage and reframe for vertical

Ask ChatGPT to find supporting shots and place them against the matching dialogue. This combines the search with a useful edit, rather than leaving you with another list of clips to arrange.

In “The Price Decision,” find our close-ups of shaping and filling the pastry. Place them above the owner’s matching explanation of the work involved, with her dialogue continuing underneath. In “The Customer Request,” show the handwritten menu when she describes the change. Return to her face for the result. Keep “The Launch That Flopped” mainly on her delivery and reaction. Preserve the approved dialogue timing in all three Shorts.

The illustration shows pastry B-roll above the interview footage, with continuous dialogue underneath and captions on a separate track.

For a separate creator example, see Matt Penny’s supporting-image and caption layers at 5:01.

If your recordings lack an establishing shot, ask ChatGPT to source stock footage or generate an illustration and place it where it helps. A generic bakery exterior can establish the subject; use the owner’s actual footage for her products and specific events. Make generated or reconstructed material clear when viewers could mistake it for the bakery itself. Media generation uses invideo credits.

With the supporting shots in place, direct the portrait compositions:

Make the three Shorts 9:16. Keep the owner comfortably framed during her answers. Include the interviewer when a question or reaction is essential. In the pastry close-ups, keep the hands and the action visible. Preserve the full menu within the portrait canvas when she points at it, rather than cropping away the relevant line. Leave space for captions.

A wide two-person shot and a close-up of working hands need different crops. For a wide menu demonstration, fitting the complete shot within the portrait canvas may be clearer than filling the screen. Check movement as well as still frames. You can refine position, scale and animated reframing manually with keyframes, or describe the specific correction to ChatGPT.

7. Add captions and finish the sound

ChatGPT can apply shared caption styling across the batch, then adjust individual moments around the compositions:

Caption the dialogue in all three Shorts using one readable font, size and colour. Use short phrases, check the bakery name and product names against the audio, and keep text clear of faces and working hands. During the menu shot, move captions away from the line she points to. Keep opening questions separate from the spoken captions so their timing remains independently editable.

In a separate creator example, The Future Of AI’s portrait edit at 7:15 shows captions with the speaker in a vertical Editor preview.

Watch once without sound to check readability and line breaks. Check that captions do not reveal the launch’s outcome before the owner says it. A shared style gives the batch consistency; placement still needs to suit each shot.

Then ask ChatGPT to finish the dialogue and music with different treatments where the stories need them:

Balance the interview dialogue and smooth abrupt audio joins across all three Shorts. Reduce distracting room noise while keeping the voices natural. Add a quiet stock music bed to “The Price Decision” and “The Customer Request.” For “The Launch That Flopped,” lower the music before the owner describes the outcome and keep her reaction clear. Fade each track at that Short’s ending. Keep the current picture cuts and captions unchanged.

Listen to every cut and ending. If cleanup makes the voice sound thin, ask for a lighter treatment on the affected passage. You can also adjust levels, fades and the mix manually. Keep the three stories distinct instead of giving every Short the same effects and music cues.

8. Export each finished Short

Play each timeline as a complete video before export. Check that the opening leads to its payoff, the captions are readable, and the ending does not depend on the next part of the interview. Make sure the three cuts offer different reasons to watch.

Export “The Price Decision,” “The Launch That Flopped” and “The Customer Request” as three separate finished 9:16 videos, using each timeline’s name for its filename.

The illustration shows H.264 MP4 selected for the finished Short. Export each named timeline separately.

You can also select each timeline and export manually. Watch the rendered files before uploading them to YouTube. Keep the editable timelines for a new hook, a revised caption or another platform format.

Editable-project export serves a different purpose: continuing the edit in Premiere Pro, Final Cut Pro or DaVinci Resolve. Specify that destination separately from your finished-video request.

Devesh Sharma’s export menu at 5:26 shows the finished-video and editable-project delivery choices.

This workflow is useful when you want help turning a long recording into several developed edits, with control over each story and finish. If you prefer giving instructions inside the editor, our built-in-assistant Shorts workflow follows that route. For the broader ChatGPT process, see editing videos with ChatGPT.

Frequently asked questions

Can ChatGPT find moments without my timestamps?

Yes. Once the footage is in the connected editor, describe the audience and the kind of moments you want. ChatGPT can search spoken content and visual action. Ask for source references and the context each passage needs, then review the selections in the recording.

Can ChatGPT make several Shorts in one request?

Yes. Identify the source, the number of cuts and what should distinguish them. Ask for separate named timelines and specify shared instructions such as caption style. You can discover candidates first when the choices are still open, or go straight to assembly when you already know the passages.

Can I start from a long video I have already edited?

Yes. Upload the rendered video or identify an existing invideo Editor project. You can also import an XML project from Premiere Pro or DaVinci Resolve with its source media to retain an editable, layered timeline. A rendered file contains the combined picture and sound; the project gives you access to its individual layers.

Does making Shorts through ChatGPT use invideo credits?

Ordinary MCP editing, including footage analysis, transcription, subtitles and audio cleanup, does not consume invideo credits. Your ChatGPT plan and usage allowance apply separately. Generating images, video or audio uses invideo credits.

Can I change the Shorts manually or continue in a new chat?

Yes. The full professional editing suite remains available, and ChatGPT can continue from your manual changes. A new conversation can read the current project, but it does not inherit the previous discussion. Bring a handover with the project link, timeline names, approved choices and remaining work.