Turn podcast highlights into vertical videos for Instagram Reels, YouTube Shorts, and TikTok with Bannerbear Workflows.
In this tutorial, we’ll build a reusable four-step workflow to join multiple clips, resize them to 1080 × 1920, burn in subtitles, and add background music with audio ducking to keep speech clear.
You’ll learn how to set up inputs, test each step, and reuse the workflow with different clips and music. We’ll also cover triggering runs through the Bannerbear API with Node.js or using an AI assistant connected to Bannerbear MCP.
Build it once, then use it for podcast episodes, interviews, webinars, and livestream highlights.
Links: ● Start a free trial: https://www.bannerbear.com/pricing/ ● How to Set Up Bannerbear MCP in Claude Code: https://www.bannerbear.com/blog/how-to-set-up-bannerbear-mcp-in-claude-code/ ● How to Set Up Bannerbear MCP in Codex: https://www.bannerbear.com/blog/how-to-set-up-bannerbear-mcp-in-codex/ ● How to Set Up Bannerbear MCP in OpenClaw: https://www.bannerbear.com/blog/how-to-set-up-bannerbear-mcp-in-openclaw
Video Transcription
[00:00:00] Hitting publish on a podcast episode seems like the finish line, but there's still more work left to do. Like any other product, a podcast needs marketing or else the people who'd enjoy it never find out it exists. Short-form video platforms like Instagram Reels, YouTube Shorts, and TikTok are a great place to do that.
Since these platforms recommend clips based on what each viewer likes to watch, your video can reach people who have never heard of your show, and it gets pushed out to even more people when viewers engage with it. In today's video, I'll show you how to automate the video editing process with a Bannerbear workflow, focusing on joining multiple clips together, resizing them to vertical, and adding subtitles and music.
Let's get into it
[00:01:00] First, let me show you what a Bannerbear workflow is. A workflow in Bannerbear lets you chain multiple video tools, images, or animations into one template. You declare the inputs it accepts, which can be text, a URL, a number, or a Boolean, and then reference those inputs inside the steps.
To follow along, you'll need a Bannerbear account, and there's a free trial link in the description if you don't have one yet. You'll also need two sample video clips and an audio track to test with. Log into your Bannerbear account and head to the Workflows tab in the sidebar. Then create a new workflow.
I'm calling mine Multi-Clip Podcast Video Editing Workflow.
[00:02:00] In the Inputs section, declare the inputs your workflow needs. These are the only things that change from workflow run to the next. We'll add three, two for video clips and one for background music. Set their types to URL and keep the required checkbox checked.
In our example, video one will be the video's hook, and video two will be a highlight from the episode, and music will be the background music With the inputs declared, we can build the editing workflow itself. Each step can take in inputs, work on them, and pass the result on to the next step. So the order of the steps is the order of the edit.
In the Steps
[00:03:00] section, select Tool from the dropdown and click Add Step. Our first tool is Join Videos. For the video URLs, click the + Ref dropdown and pick video one and video two. So the step joins the two input videos into a single clip. Add a second step and choose Resize Video. This time, reference the previous step's output as the video URL, since we want to resize the joined clip and not the two originals.
Set the width and height to 1080 by 1920 pixels so the clip comes out as a vertical video. The third step is Transcribe and Burn Subtitles. Reference the previous step's output as the video URL again so that the captions are added to the vertical
[00:04:00] clip. Then set the other fields to suit your show, like words per segment, font size, and font style.
The fourth step is Add or Replace Audio. Reference the previous step's output as the video URL and the music input as the audio URL. Set the mode to Mix, and set ducking, which is how much the music drops when someone speaks too heavy, so that the host voice is still audible. Once everything is configured, hit Save Workflow.
Now, let's test the workflow and make sure every step does what we expect. On your workflow's overview page, upload your video and music files or fill in the form with their links.
[00:05:00] Then click Send Request to trigger a test run.
When the run finishes, you can click View Log to see the result. The result from each step is listed on the page, and you can click on each one to view it separately. And this is the final result
You can also keep using the same form to trigger the workflow. It's a good fit for any repetitive task where the inputs change every time, but the job doesn't happen often enough to justify a full automation, like creating clips for your weekly podcast episode. Filling in a form works well for a weekly job, but some jobs are reoccurring enough to be worth automating.
For example,
[00:06:00] a news or daily podcast that needs clips every day, a production team that produces clips for multiple shows, or the same highlight that needs to be edited several times with a different intro or music track for each platform or campaign. For these jobs, we can integrate the workflow into an automation script through the Bannerbear API.
Here's a short Node.js script that sends a post request to the workflow runs endpoint, which is the endpoint that triggers a workflow run. In the request body, we pass the workflow UID along with an input object that holds our three links. The input names have to match the ones we declared in the workflow, or else Bannerbear won't know which link goes where.
The API runs asynchronously,
[00:07:00] so the first response only gives us the run's UID and its status. To get the finished video, we poll the run until it completes and returns the final result. Bannerbear also has an MCP server which you can connect to your AI assistant. Once it's connected, you can trigger a run just by asking for it in plain language.
For example, I can type, "Run my multi-clip podcast video editing workflow with this file as video one, this file as video two, and this file as music." The AI assistant calls the tools needed and hands back the finished video URL in the same conversation If you'd like to set this up yourself, we have dedicated guides for Claude Code, Codex,
[00:08:00] and OpenClaude, and they're all linked in the description.
So that's a podcast clip workflow you can build once and trigger any time or even automate. The same chain works for other long recordings too, like interviews, webinars, or live streams you want cut into vertical clips. This four-step workflow is a simple demonstration of how a workflow chain steps together, and Bannerbear's tool library covers a lot more that you can try in the same workflow.
For example, you could add trim video to cut a clip down to the best moment of the conversation before joining it with the rest. Add an image overlay to place your podcast logo in the corner of the video, or create a GIF preview that you can drop into a newsletter announcing a new
[00:09:00] episode. If you'd like to watch more tutorials like this, subscribe and turn on the notification bell.
I'll see you in the next one.



