Paste the YouTube URL. Juma reads the transcript, auto-extracts timestamped chapters, writes a hook line, summary, CTA block with links, and a hashtag set.
Paste a public YouTube URL and Juma starts with the transcript, not a placeholder paragraph. Timestamped chapters come straight from the transcript markers, the hook line is anchored in a striking moment from the video, and the full description includes a summary, CTA block with links, and a hashtag set.
Chapters are the part viewers actually use to navigate, and they usually get skipped because writing them by hand means scrubbing the timeline for transition points.
The description holds the watch; the title gets the click. Juma writes ten title variants for the same video in one continuation prompt, each tagged by hook angle, with character counts and a recommended A/B test pick. The angle tagging spans curiosity, transformation, how-to, listicle, contrarian, and urgency, so the team can pair the two that target the widest spread of viewer emotion. The transcript and the description Juma just wrote stay in context, so the variants stay grounded in what the video actually delivers, not a guess at the topic.
Now write 10 YouTube title variants for the same video, tagged by hook angle, with character counts and a recommended A/B test pick.
Chapters are the part of a description that takes longest to write manually and the part viewers use most. For videos already live where the team only needs to add chapter markers, Juma pulls timestamps directly from the transcript and returns a clean chapter list ready to paste into the existing description. Each chapter starts on an actual content shift in the transcript, not an approximation, and the chapter titles describe the segment in three to seven words. The output can be plain text, emoji-prefixed, or bold-sectioned, depending on the channel's existing style.
For this YouTube video, generate just the chapters from the transcript. Each chapter title should be 3 to 7 words and start on an actual content shift, not a fixed interval.
YouTube treats tags and hashtags as separate fields with separate purposes. Tags help the algorithm classify the video; hashtags help viewers discover related content through hashtag search. Juma reads the transcript and proposes both: ten to fifteen tags ranked by relevance and search volume, plus five to seven hashtags balanced between broad (high volume, high competition) and niche (lower volume, higher intent). The output explains the reasoning for each pick so the team can adjust based on channel-specific signals, like what's already converting in YouTube Studio.
For this YouTube video, propose 10 to 15 tags ranked by relevance and 5 to 7 hashtags balanced between broad and niche. Explain the reasoning for each pick.
For channels that publish localized versions across regions, Juma translates the description into the target language and adapts the cultural references, idioms, and CTAs at the same time. Direct translation flattens marketing copy. The output preserves the structure (hook, summary, chapters, CTA, hashtags), translates the chapter titles, and rewrites any culturally specific phrasing for the new audience. The team names the target languages and Juma returns one description per language, ready to paste into the regional channel's video page.
Translate the YouTube description into Spanish, French, and German for Notion's regional channels. Preserve the structure, translate the chapter titles, and adapt any culturally specific phrasing or CTAs.
For agencies managing a client channel or in-house teams clearing a back catalogue, doing one description at a time isn't viable. Paste a list of YouTube URLs (up to ten in one session) and Juma writes a full description for each, delivered as a single CSV with one row per video. Each row covers the URL, hook line, summary, chapters (with timestamps), CTA block, hashtags, and the recommended tags. The CSV format makes it easy to filter to only the videos that need new descriptions, sort by publication date, and queue the approved descriptions for YouTube Studio bulk-edit.
Paste a list of 10 YouTube URLs for our client Notion. For each, write a full description: hook, summary, chapters, CTA block, hashtags. Deliver as a single CSV.
A Juma Project is a shared space where the team stores everything Juma needs to know about a channel. Create one project per channel (or per client, if the agency manages multiple), add context as the team learns more, and Juma uses what's relevant every time a flow runs. For descriptions, this is what keeps the CTA block consistent across hundreds of videos and the tone aligned to the channel's existing voice.
Channel Voice Guide
How the channel sounds: formal or conversational, slang or buttoned-up, short paragraphs or long. With this in the project, descriptions come back in the channel's tone instead of a generic YouTube voice.
CTA / Link Block
The standard subscribe text, sponsor and affiliate links, lead magnet URLs, and social handles the team appends to every description. With this in the project, Juma drops it in verbatim, so the CTA block stays consistent across every video the team publishes.
Past Descriptions
Three to five descriptions from videos the team is proud of. Juma uses them to match the channel's structural conventions: how the hook is phrased, whether chapters use emoji prefixes or plain text, what hashtags consistently appear, where CTAs sit relative to the summary.
Brand Guidelines
Voice, banned phrases, hashtag conventions, naming for product features. Keeps every description aligned to brand standards without manual review for every release.
Add a short description to each knowledge item in the project's info field so Juma knows what each file contains and when to use it. For example:
Juma reads the video transcript along with the embedded timestamps and detects natural content shifts: topic changes, speaker transitions, scene cuts that map to a new section. Each shift becomes a chapter marker, and the chapter title is a three to seven word summary of what that segment covers. The output is ready to paste into the video's description on YouTube.
For chapters to render on YouTube, the description must start with a 0:00 timestamp and contain at least three chapter markers spaced at least ten seconds apart. Juma follows that format by default. For very short videos (under 90 seconds) where chapters aren't useful, Juma flags that the chapter block can be omitted and writes a cleaner short-form description instead.
A full description covers six components: a hook line written to stop the scroll, an 80 to 100 word summary that previews the video's value, timestamped chapters auto-extracted from the transcript, a CTA block with subscribe text and key links, a hashtag set balanced between broad and niche, and an optional tag suggestion list for YouTube Studio's separate tags field.
The hook is anchored to a striking moment in the video when the dialogue supports it, or to the announcement's framing when the video is cinematic and music-led. The summary previews what viewers will see without giving away the payoff. The CTA block is dropped in verbatim from the channel project when one is set up, so it stays consistent across every video the team publishes.
Yes, when the team adds a Channel Voice Guide and three to five past descriptions to the project. With those in place, the new description matches the structural conventions the channel has already established: how the hook is phrased, whether chapters use emoji prefixes or plain text, where the CTA block sits relative to the summary, and what hashtags consistently appear.
Without past descriptions in the project, Juma defaults to a clean YouTube format: hook, summary, chapters in plain text with timestamps, CTA block, hashtags. That's fine for channels still establishing voice. For channels with consistent existing conventions, loading the past descriptions is what changes generic output into channel-native output.
Yes. For long-form videos (30 minutes and up) and podcast episodes uploaded to YouTube, chapter extraction is more useful, not less. The transcript is denser, the content shifts are clearer, and viewers rely on chapters more heavily to navigate the runtime.
For videos over an hour, Juma typically returns eight to fifteen chapters depending on how the conversation is structured. For shorter clips and product demos, three to six chapters is normal. The chapter titles stay descriptive of the segment rather than teaserish, because chapter titles serve navigation, not click-through. If the team wants more curiosity-driven chapter titles for a specific video, they can ask in the prompt and Juma adjusts.
Yes. YouTube allows up to 5,000 characters in the description, but only the first 100 to 150 characters appear above the "Show more" fold on most viewers' screens. Juma puts the hook line and the primary keyword in the first 100 characters so the visible-by-default copy carries weight. The full description stays well under the 5,000 character limit even with chapters, CTAs, and hashtags included.
SEO best practices are handled implicitly: the primary keyword appears in the first sentence and again in the summary, hashtags include the keyword as one of the tagged entities, and the chapter titles avoid keyword stuffing while still being descriptive. Human review on every output: the team adjusts the keyword targeting and CTA block based on the channel's strategy.
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.
