IdeaFORGE Turns One Source Story into Podcasts, Slideshows and Videos

Producers set up a build with the settings that fit the format they pick. In the web console, each format shows its own length options, and the estimate appears before you launch. A Test build option applies a preview preset for the chosen format, and turning it off returns your own settings. Test build is not offered when rebuilding, for formats the preset would not change, or when creating a story's first build.

Producers working from long stories and research can use that text in slide scripts and in the hooks and scripts of reels. A much higher source limit applies to builds started from the console or by API, and shorter sources need no change. Video sales scripts and long-form podcast planning keep their current limit.

Podcast producers revise one segment of a sponsored episode without rerunning the build. Segments are regenerated individually from the console's segment editor or by callers on the API, matched to the episode's sponsor reads, and a request that fails the up-front checks is declined before any spend. Sponsor reads that play from a pre-recorded file cannot be regenerated, some older sponsored builds need a rebuild first, and in series-backed builds a regenerated segment may use the default voice.

Brand teams start a new style from one that already works. Clone, in the editor of an existing style, copies the name, description, prompt and links to the reference images into a new style that is not the default. Integrators and agents clone styles the same way, and Clone asks you to save unsaved edits first.

Audiobook and deck producers hand off completed files that carry their cover art and title. Re-stitching an audiobook that has cover art keeps it in the M4B and MP3 files, and PDF slideshows carry a document title. PPTX and PDF exports stop with a message naming any unreadable slide image. Existing audiobooks and decks keep their current files until they are re-stitched.

Marketing teams run campaigns at scale with the style, voice and template they pick for each entry. An entry's template takes priority over the campaign default when the template fits the entry's format, and importing names the entry and setting that needs attention. Builds made from a series also apply the series' host and guest voices and intro settings, and a voice picked in the build takes priority. These are changes to existing behavior: output and cost can differ for campaigns with a default template or per-entry style or voice and for series-backed builds, and entries that point to a deleted or other-workspace item fail with an error.

Producers recover a failed or cancelled deck, audio or video build whose generation steps completed with a re-stitch from the console or by API, and a build with missing steps is refused with those steps named. Batch retry restarts the failed (default), cancelled or parked builds of a bulk batch through the API and agent tools, skips builds that cannot rerun, and returns a per-item list of what was accepted. Wait for a build's re-stitch to finish before retrying it.

Teams planning media builds see live-action, research and voice spend counted in the cost estimate and the Usage page. Live-action clips and keyframes, research calls and voice synthesis from other voice providers record their spend, and members with read access can run cost estimates. This changes existing behavior for workspaces running those capabilities: a build that would exceed a hard limit can be parked or refused. Totals may omit some tool usage, and limits are set by the platform administrator, not by workspaces.

Members invited by a workspace owner remain pending until they accept from the workspace switcher, and a user-bound API key acts with the lower of its role and the member's current role. Sponsors, music and brand styles resolve inside the workspace running the build, and a new reference to another workspace's item is declined up front, naming the item, for accounts with more than one workspace. These apply to workspaces that use sign-in accounts with roles, and workspace owners send invitations by API.

Agent builders get the information they need to supervise builds and bring files in. Over the API, a build read reports render warnings, and create and rebuild requests report an approximate queue position. Agents that can send an HTTP upload register an asset with a checksum and upload it directly, so incomplete uploads are refused. For teams that bring their own provider keys, a bulk request reports the items that lack a workspace key and rejects those items while the rest are accepted, covering the workspace's default voice, and campaigns can run in those workspaces.

Editors and reviewers who work from the keyboard close stacked windows one layer at a time and sort tables from their column titles. Sorted columns are announced, and action menus respond to arrow keys, Home, End and Escape. In workspaces that use sign-in accounts, each browser tab keeps its own workspace, and a signed-out tab returns to sign-in and brings you back to where you were. Leaving the story page with an unsaved edit to the story body asks first, including on reload.