Skip to content

Studio

The Studio category holds the long-form and planning generators. They are all streaming calls, so text appears as the model writes it. Long pieces can take tens of minutes on CPU.

Planning

Content Strategy

Studio → Content Strategy calls InsightsService.GenerateContentStrategy with an objective (for example “Thought Leadership” or “Lead Gen”), channels, and an audience. It returns content pillars, each with example topics and a recommended share of the mix, a posting cadence, a 0–100 alignment score, and a summary.

Strategy Roadmap

Studio → Strategy Roadmap calls GenerateStrategyRoadmap with an objective and a planning horizon in weeks. It returns phases (for example “Foundation”, “Momentum”, “Scale”), each with a timeframe, initiatives, projected momentum, and an outcome, plus a 0–100 readiness score.

Both are grounded in your diagnosis, your published content, and live market signals, and both run through the judging ladder. A pillar or phase the judge disagrees with is withheld.

Long-form writing

ScreenCallYou provideYou get
WhitepapersGenerateWhitepaperTopic, industry, length (short, medium, in-depth)A long-form document
PodcastsGeneratePodcastSource material, host style (conversational, deep_dive, news), target minutesA podcast script
ShortsGenerateShortsTopic, platform (youtube, instagram), visual styleA short-video script with visual cues and on-screen text
PresentationsGeneratePresentationObjective, slide count, brand colorsA slide-by-slide outline
InfographicsGenerateInfographicData source, layout (timeline, comparison, process)Infographic copy: title, data points, icon ideas
ReportsGenerateReportTimeframe, metrics to highlight, export formatA report draft

These screens produce text. The podcast and shorts screens write scripts; they do not record audio or render video. A report is written by the model from the timeframe and metric names you give it. It does not read your metrics, and the export format is a hint in the prompt, not a file conversion.

The whitepaper generator is the most grounded. It uses the long-form copywriter agent, injects the same diagnosis context that content generation uses, and says so when it has none instead of inventing facts. Its length setting sets a real token budget.

To approve or schedule a Studio piece, save it as content with ContentService.SaveGeneratedContent.

Audio and video from the background engines

Finished audio and video come from background engines, not from the Studio screens. They run on active campaigns:

EngineOutputHow it publishes
Podcast engineEpisodes, scripted and synthesized to WAVStaged for review; published by inclusion in the podcast feed
Audiobook engineA weekly long-form bookThe same feed, as its own channel
Shorts engineShort videosStaged as a YouTube upload. Refuses to start when the image has no video tool.
Document engineDocumentsDaily, for review

Speech is synthesized inside PULSE (piper). Publishing each of these depends on its channel being armed. See Publishing controls.

Courses and avatars

Content → Courses calls CourseService.GenerateCourse with a topic and a number of modules. For each module the model writes a lesson title, a teaching script, and a quiz grounded in that script.

If you pass an avatar and the deployment can generate video, each module also gets a video. If it cannot, the module is saved with no video, not a placeholder, and the server logs why.

Avatars come from AvatarService.CreateAvatar: a name, a photo, and optionally a voice sample, both base64-encoded.