We've redesigned the new episode input field to align more closely with common workflows in other AI applications like ChatGPT and Perplexity. Starting today, you can provide additional instructions via a "user prompt" to steer the output of your podcast episode.
Background
Previously, anything that was added into the new episode text box was considered a source, including text. Based on user feedback, we realized there are often two distinct input types used when creating a podcast episode.
The first type are sources - individual files, websites, etc. that provide context for the episode. These are foundational documents that provide up-to-date information to ground the podcast generation in user-provided knowledge.
Jellypod supports over 70 different source types, from Powerpoint, PDFs, and Word documents, to website blogs, YouTube links, and more.
The other input type is a prompt - a way to steer the episode towards specific details, topics, or a conversation style. Until today, the recommended approach was to add in a "text source" that provided additional guidance for the AI model. But now we fully support the inclusion of an optional user prompt that provides additional instruction on how the episode should be created.
How it Works
The new episode flow remains extremely similar. To create an episode, you can still drag and drop sources into the input field which will be processed and used as context (or add a website link).
However, just like in ChatGPT, short text inputs will remain in the text box rather than being added as a source. This becomes your "user prompt", which can be used to guide the episode creation.
We recommend adding all of your sources into the input box first, and then optionally type a few words to steer the output.
Once you've added all your sources and additional instructions via your user prompt, click Create Episode. Jellypod then processes it all together and creates a rich episode outline for your review.
Try it out today and let us know what you think!