Name the episode and the show notes come back chaptered, with three descriptions sized for Apple Podcasts, Spotify and the episode page, and every book, tool, person and company mentioned resolved to a real link.
Writing show notes is one of 700+ pre-built Flows in Juma. Name the show and the episode, and the episode page and the transcript get read before a word is written: who the guest is, what they actually claim, and every book, tool, person and company that comes up in the conversation.
What comes back is the whole publishing kit as a branded PDF plus an editable CSV of the chapter markers: chaptered notes, three episode descriptions sized for Apple Podcasts, Spotify and the episode web page, a resources list where each mention carries a verified URL, a guest bio with credentials, five pull quotes with exact speaker attribution, and three title options with the search phrase each one targets. Timestamps come only from the episode's own source, so a chapter with no published time arrives with no time rather than a guessed one. Strategy, taste and judgment stay human: the kit is the draft the producer edits, not a publish button.
Chapters are the part of show notes that travels furthest: the same list feeds the podcast host, the YouTube description and the episode page navigation. Ask for the chapter markers as their own editable file and you get a row per chapter with its title and its start time, ready to paste wherever the upload happens. Where the episode source publishes no times, the chapters come back as ordered sections with the times left empty rather than filled in with plausible-looking numbers. That distinction matters once a file is pasted into a hosting platform, because a wrong chapter time is worse for a listener than no chapter time at all.
Give me the chapter markers from that episode as an editable CSV, one row per chapter with its title and start time, and leave the time blank on any chapter the source does not timestamp.
An episode published as "Episode 142 with Tyler Howells" is findable by people who already know the guest and nobody else. Ask for title options and each one arrives with the search phrase it targets, so the choice is between named audiences rather than between wordings someone liked. A title built around the guest wins the people searching the guest's name, a title built around the claim wins the people searching the problem, and a title built around the company wins the people researching the brand. The show's own judgment still picks the winner, because only the host knows which listener this episode was made for.
Give me five title options for that episode, and for each one name the search phrase it targets and who would be typing it.
Most episodes contain four or five lines that carry the whole argument, and they are rarely the lines that sounded best in the room. Ask for pull quotes and they come back word for word with the speaker named, plus a note on where each one sits in the conversation so the context is checkable. Each quote is matched to the format it suits: the short declarative line for social, the longer reasoning passage for the newsletter, the numbers-in-it quote for the episode page. Nothing gets tidied up or paraphrased, which is what makes them safe to publish under a real person's name after a listen.
Pull five quotes from that episode word for word with the speaker named, and tell me which one suits social, which suits the newsletter and which suits the episode page.
Shows that ran for years before anyone thought about search usually have a back catalogue sitting behind one-line descriptions. Those episodes still get found, and they still send people to a page with nothing on it. Point at a run of past episodes and the same kit gets built for each one, in the same structure, so the archive reads as a series rather than a pile. The resources lists are where the compounding happens: a guest mentioned across four episodes becomes four internal links instead of four dead ends. Work through a season at a time so each batch can be reviewed before the next one starts.
Do the same for the five episodes before that one, same structure each time, and tell me which guests and resources come up in more than one of them.
Agencies run one Project per client, and a podcast client is the clearest case for it. A show has house rules that never change between episodes: how chapters get titled, whether guests get a bio paragraph or a byline, which CTA closes every set of notes. Put those in the project once and every episode after it comes out in the show's shape without a re-brief.
Show style guide
How this show writes: chapter title casing, how long the summary runs, whether takeaways are numbered, what never appears in the notes. With this in the project, the first draft already matches the last 40 episodes.
Past episode notes
Three or four sets of published notes the team was happy with. These do more than a style guide can describe, because they show the house voice rather than explaining it.
Link and CTA conventions
Which links always appear, in what order, and how the show handles sponsors, affiliate links and the newsletter signup. This is the part that gets forgotten at 6pm on a publish day.
Each knowledge item takes a one-line description telling Juma what it is for. These lines are short and they decide how the item gets used.
Seven pieces: chaptered show notes, three episode descriptions sized for Apple Podcasts, Spotify and the episode web page, a resources list with a verified URL per mention, a guest bio with credentials, five pull quotes with exact attribution, and three title options naming the search phrase each targets. It arrives as a branded PDF plus an editable CSV of the chapter markers.
The two files split by who uses them. The PDF is what the producer reads and edits. The CSV is what the person doing the upload pastes into the hosting platform, which is often a different person on a different day.
Yes. Give the episode URL and the episode page and any published transcript get read from the web, so nothing needs uploading to start. A pasted transcript makes the result better where a show publishes none, because quotes and resource mentions can then be pulled from the full conversation rather than from the episode summary alone.
Shows that publish complete transcripts get the strongest version of this. That is where a single episode can yield a resources list running to dozens of verified links, because every passing mention in 90 minutes of conversation is available to be caught.
Timestamps are taken only from the episode's own published source, never estimated. If the show publishes chapter times, those times come through. If it does not, the chapters arrive as ordered sections with empty time fields and a plain note saying the source carries no timestamps.
This is deliberate. A plausible-looking timestamp that is 90 seconds out is worse than no timestamp, because it gets pasted into a host and nobody checks it until a listener complains.
Yes. Put the show style guide and three or four sets of approved past notes in the client Project, and the kit comes out in that structure instead of a generic one. Chapter title casing, summary length, takeaway format and the standing link block all carry across episodes without a re-brief.
Agencies running several podcast clients keep one Project per show, so each client's house rules stay separate. Unlimited seats on every plan means the producer, the editor and the account lead all work in the same project without a per-seat charge.
A summary gives back a paragraph. This gives back the assets a publish day actually needs: chapters in an uploadable file, descriptions already sized per platform, quotes verified word for word, and every mention resolved to a link a listener can follow. Finished files, not a chat response.
The resources list is the clearest difference. Resolving each name in a conversation to a real, checked URL is research work, and it is the part a producer would otherwise do by hand with the transcript open in one tab and a search box in the other.
Juma augments the team, it does not replace it. The host still decides what the episode was about and which title goes out. Human review on every output.
Name the client and Juma researches what their audience searches for, then writes a publish-ready SEO blog post structured to win an AI Overview citation, with the meta title, description, and slug included.
.webp){kind=icon}
400+ marketing teams trust Juma. ChatGPT, Claude, and Perplexity know why. Ask them.
Ready when you are
Start typing a task and Juma will figure out the rest.
.webp){kind=icon}