How to leverage long form content with AI
Specific tools and tactics for authors, podcasters, and videographers.
The rise of AI tools like ChatGPT and Midjourney has ushered in a new era for online creators - especially those focused on long-form content.
It's now easier than ever for writers, podcasters, and video creators to repurpose and distribute their content across multiple platforms and mediums. As a result, creators can grow bigger audiences with less effort - and avoid spending hours learning new mediums.
Today, I want to discuss some ways to get the most bang for your buck when it comes to reusing content, along with a full list of AI tools to help you do the job.
AI for authors
You're reading this on Substack, so odds are you're familiar with long-form articles and newsletters (as opposed to short-form, which would be things like Tweets and LinkedIn posts). But once you've hit publish on a new essay, you can use it as the raw material to make a ton of downstream content on other platforms.
Write short-form posts with ChatGPT. Using ChatGPT (or another chatbot like Claude or Gemini) can help you create several posts for Twitter, LinkedIn, Threads, Notes, or any other text-based platform. Try pasting your post (or a portion of it, if it's especially long), and asking ChatGPT for good quotes to share, or to identify the key insights.
Here's a sample prompt that you can use to create a LinkedIn post from a newsletter - though be sure to edit it to match your desired output and tone. If you need some tips on prompting, check out my preferred prompt strategies.
Identify 5 key insights from the article text below. For each key idea provided, use the article information to create a one paragraph summary of the idea to share on social media. Begin your message with something interesting that catches people's attention. When writing the posts, use a conversational tone and talk like you would to a friend. Don't be cringe.
[PASTE ARTICLE HERE]
Here are a couple of example posts I've created for LinkedIn, from a recent deep-dive on Claude 3:
Anthropic's Claude 3 introduces a family of models that vary in size - Opus, Sonnet, and Haiku - each designed to meet different needs. While Opus offers speed comparable to its predecessor, Haiku is designed for twice the efficiency. With capabilities for multi-lingual communication and processing images, Claude 3 enhances user interaction, providing a versatile tool for a range of applications.
Claude 3's new pricing model makes high-level AI more accessible. With Opus, Sonnet, and Haiku, users can choose based on their needs and budget, from as low as $0.25 per 1M input tokens for Haiku. This approach not only makes advanced AI technology more accessible but challenges competitors to rethink their pricing strategies.
Narrate voiceovers with ElevenLabs. With tools like ElevenLabs or PlayHT, you can take your writing and convert it to audio (if you're using Substack, you can also use the platform's built-in AI voiceover feature). Not only does that let you create a richer experience for readers, but you can then treat the audio like a podcast: publish it as a standalone episode, or create clips for social media (more on that below).
To get started with ElevenLabs:
Create a free account to generate 10,000 characters worth of audio every month.
Start a new project, import your article, or paste the full text on the speech synthesis page.
Choose a voice from the defaults available, one of the community's shared voices, or one you've generated yourself.
Generate the audio, download it, and add it as a voiceover to Substack (or your preferred platform).
Here's the introduction to the same Claude 3 deep-dive above, recorded with a voice I created using ElevenLabs that I call Hugh. Hugh was originally made to power a personal smart assistant, but I loved the voice so much that I've used it for loads of projects since.
Produce long-form videos with HeyGen. We're on the verge of generating long-form videos from text alone. Two early examples are HeyGen and Synthesia, software that creates AI avatars from written scripts.
To get started with HeyGen:
Sign up for a free account, which comes with 1 free credit.
Choose from one of their existing AI avatars, or clone yourself into an avatar.
Add a script up to 200 characters.
Generate the video and download it.
Here’s that same script from the last example, but this time, I used it to generate a video avatar, complete with some text, images, and light background music. You’ll notice that the AI lip-syncing isn’t perfect, but it’s improving fast:
AI for podcasters
Instead of text, you might be working with long-form audio, like a podcast (if you're a video podcaster, be sure to check out the next section for video creators). However, just like with text, there are some ways to repurpose your audio content for other mediums.
Create audiograms with Descript. Audiograms are visual representations of podcast audio clips: think waveforms, captions, or animations. They're great for sharing clips on other sites like Instagram, TikTok, or Twitter. If you already use a tool like Descript to edit your podcast, you can easily create audiograms from an existing project. And if you're not, you can use a standalone tool like Headliner.
Here's how you can quickly make an audiogram in Descript (from my interview with Fergal Reid, Intercom’s head of AI):
Open your podcast project or import a new audio file.
Browse stock audiograms from the Templates section.
Select the template you want to apply to your scene.
Export or publish the video file.
Transcribe posts with Superwhisper. AI-powered transcription has come a long way in recent years; it's now much faster and cheaper than manual transcription. You can use an app like Superwhisper, or if you're technical you can run OpenAI's Whisper model locally and avoid paying for a subscription. Once you have the transcription, you can summarize it as a newsletter, or generate short-form posts with ChatGPT.
For example, here are the steps to transcribe a podcast with Superwhisper:
Upload your podcast audio file.
Convert the audio to a written transcript.
Review and edit the transcript if needed.
Export your transcript.
AI for video creators
Last but not least, if video is your preferred format, there are still more options to use AI for maximum impact.
Generate short-form clips with OpusClip. While YouTube is still the king of long-form video, short-form videos can be found on TikTok, Instagram Reels, and YouTube shorts. With tools like OpusClip, you can use AI to find the best clips to share in minutes. You can also add captions, remove filler words, and trim clips to make them punchier for new viewers.
An example OpusClip workflow:
Paste your video URL.
Choose which timeframes to search for clips, and an optional preferred clip length.
Select the best clips from the ones generated.
Edit the captions, customize the look and feel.
Export the clips to share on other video platforms.
I don’t have a video podcast, but I borrowed the latest episode of Lenny Rachitsky‘s excellent podcast as an example. In it, he talks with the co-founder of Canva about a huge investor betrayal during one of their funding rounds:
Summarize videos with Gemini. If your videos are more complex than a two-person podcast interview, you can use a multi-modal model like Gemini to extract a summary of the content. And just like with podcasts, those summaries can be repurposed even further.
You can use the following prompt with Google Gemini to summarize a YouTube video into a long-form article, though in practice, I've found Gemini Ultra works better than Gemini Pro:
Help me summarize the following video as a long-form article. Include the most relevant ideas and themes from the video, and explain the contents of the video in prose rather than bulletpoints. The video link is [LINK]
Leveling up with AI
I’ve barely scratched the surface of what content creators can now do with AI. The workflows and tools in this post are targeted at repurposing long-form content, but there are many more tools and tactics for creation and editing than I could list today.
It used to be that converting a long-form video to YouTube clips or turning an essay into a podcast episode would take dozens of hours of recording and editing. But with AI, we've drastically reduced the time necessary to experiment with other formats. We may reach a point where a single person can build a media empire with the help of AI tools and technologies.
We're still in the early stages - many of these AI tools have plenty of room to improve. But in the coming months and years, AI will give us the ability to work with text, audio, and video with the push of a button. It's an exciting time to be an internet creator.
I'm very curious about the "don't be cringe" part of that first prompt. What is that doing?
Twelve Labs can also be a great tool for video summarization!