VIDYA, long-form video for your learning series

One finished video per episode, made from the script you approved

VIDYA leads your Learning Series team, at Learning Series. Each episode starts as a script grounded in your Knowledge Hub. From that script VIDYA makes one finished long-form video: the presenter, a branded slide card for each slide, and your intro and end plates, in module order.

The same video does three jobs. You post it to YouTube or LinkedIn as long-form social content, you show it as a product demo, and your website plays it on the episode's public page. Nothing is posted or published on its own: every route out is a choice you make per episode.

The video features carry a Beta badge where they appear. They work end to end; the badge says they are new and may still change.

VIDYA's team sits under CREA

VIDYA is a lead under CREA on the Content team, so your learning series is content work and follows the Content room's rules. VIDYA's room, VIDYA team, works inside the Content room, and VIDYA works through three seats of its own:

Each seat starts at level L3 and runs its own clicks as agent actions, priced against its own budget. The presenter on screen is none of them: it is you, or a persona from ALPHA (next sections). Today the Learning series room reaches only the workspaces Nagent sets it up for.

Pick who presents the series once, in its settings

Open the series and find Presenter. There are three choices:

The two AI choices need a presenter. Pick it in the AI presenter field below the choices. The list shows your own personas that have a long-form presenter, then Nagent's agents, each marked Nagent agent with a Beta tag. Until you pick one the field says "No presenter picked".

Nagent's agents can present your series

Nagent can set up a presenter for each of its agents, with that agent's own face and voice. Once Nagent has set an agent up and it is ready, it is in every workspace's AI presenter list, marked Nagent agent: pick it, save, and record. There is nothing to make, approve or set up in ALPHA, and the same agent can present your series and another workspace's. An agent that is not set up yet is simply not in the list. The co-host session, the rendered clips and their credits are your workspace's own, as with a presenter you made.

A long-form presenter of your own is a persona from ALPHA, made once

The face and voice that present your series belong to a workspace persona in ALPHA, the video studio, at Personas. The same persona can present a learning series, a demo or an ad, so your presenter looks and sounds the same everywhere. See ALPHA personas and avatars for how a persona is made, voiced and approved.

1. Create a persona, or open one that is still a draft and click Complete.
2. Under Long-form presenter, click Pick a face and voice. Choose a Face and a Voice, then click Use this face and voice. Each list has stock faces and voices, and any that Nagent has set up for your workspace alone, marked This workspace's own.
3. Click Ask for approval. Another admin of your workspace, or Nagent, approves the persona, and the face and voice are approved with it.
4. Once the persona is active, click Make the presenter. The section shows the presenter's state; wait for it to read ready.
5. Back in the series settings, pick the persona in AI presenter and save.

A face or voice of your own that belongs to a real person is marked as needing a signed release, and the presenter is refused until the persona has a written release on file; for a voice, the release must cover voice. A stock face and voice need no release.

A digital twin cannot be a long-form presenter; the section does not appear on a twin. Make a brand spokesperson persona for the presenter instead.

Every episode moves through four steps

Open an episode from its series. The episode page has one step strip: Script, Presenter, Compose and Share, with one section per step below it. The strip marks the step the episode is on.

Script

The episode's title, theme, excerpt and its 12 module titles. Click Generate shoot kit to draft about 16 slides and a talking points cheatsheet. The page says what it costs (about $0.20) and how long it takes (60 to 120 seconds); keep the tab open while it runs. Regenerate drafts a fresh kit.

The Sources panel lists every claim in the script with the document and passage it came from. A workspace claim without a supporting source holds the episode until you Recheck claims after a fix, or Mark general.

Presenter

What this step shows depends on the series' presenter.

Human. Click Open studio and record. The studio shows the slides and the teleprompter side by side; use the arrow keys to move between slides and scroll the teleprompter, and space to auto-scroll. Record in the browser or upload a take, review your take, then click Finalize & upload. Back on the episode page, review the recording, trim it and edit its timeline.

AI presenter. Click Render with AI presenter. VIDYA asks ALPHA for one clip per module, read by the series' persona from the approved script. Each module shows its own state: not asked, rendering, stored or failed. Clips arrive one at a time and the page checks every 30 seconds. A failed module shows its reason and a Render again button that renders that module alone; the others are kept. When every module reads stored, the page says "Every module has its clip. Make the video in Compose."

If you edit the script after the clips were asked for, the changed modules are marked script changed and the video waits for them. Click Render changed modules to render those modules, and any failed or missing one, again. Render every module again renders the whole episode again from the script as it stands: it asks you to confirm, each clip is charged again, and the new clips replace the ones you have.

Human with AI co-host. Open the studio. The AI co-host sits beside your camera:

1. Put on headphones, so the co-host hears you and not itself.
2. Click Start the co-host. A timer shows how much of the session is left.
3. Click Record, and talk. After each of your turns the co-host shows Listening, Thinking, then Speaking, and answers out loud. The conversation is written out below the two videos.
4. Click Stop recording, then End the session.

The co-host answers from your Knowledge Hub and the episode, under the same claim rules as the writers. When it does not have a source for something, it says it would rather check your notes first than say it on camera. A session ends by itself after 30 minutes.

The recording holds both sides, side by side, with both voices. From here it is a normal human recording: review, trim and compose it as above.

Compose

Finished video shows what the video will be made of. Click Make the video. It renders in the background; a long episode takes a while, and the page checks every 20 seconds. The video holds:

• your intro plate;
• the presenter footage for each module, in module order;
• a slide card for each slide, in your workspace's brand colour, logo and name, with the slide's generated image where it has one;
• your end plate.

Your trims, skipped ranges and timeline edits from the Presenter step are kept. Make the video again renders a new one after a change; the earlier video keeps playing until the new one is ready. A published episode is refused here, because republishing makes its video again.

Compose waits for its input: a human series needs the recording uploaded first, and an AI series needs every module rendered first.

Share

Sharing opens once the finished video is ready. You choose any mix of three routes, and none of them happens on its own.

Download. Download MP4 gives you the finished file. Next to it is the pack: a title, a description with chapter lines from the module timestamps, a thumbnail and tags, ready to paste into any platform.

Post to YouTube or LinkedIn. Each button creates a draft post in MOXA with the video attached. Nothing is sent from the episode page: the post waits in MOXA for its checks and a second person's approval, like every post MOXA makes. The episode lists its posts and their state.

• LinkedIn takes a video up to ten minutes long. A longer episode is posted to LinkedIn as a link post, marked (link post) on the button.
• A video larger than the upload path takes (100 MB today) cannot be posted directly. The episode says so and points you to the download and the pack, so you can upload it yourself.

Cut-downs (Beta). Below the share routes, Cut-downs makes short clips from the finished video for social channels. Click Make cut-downs. It makes:

• an upright 9:16 short for each chapter, up to a minute long, cut from the chapter's best moment, with the chapter's title on screen;
• a 16:9 teaser of up to a minute, cut from the episode's best moment.

The best moments come from reading the finished video. When the finished video is ready, it is added to your Knowledge Hub and read there. If the reading is still under way when you click Make cut-downs, the panel says it is waiting for the video to be read, and the cut-downs start by themselves once it is done. When a moment cannot be picked, because the reading failed, VIDYA was out of budget or the wait passed six hours, the clip is cut from its chapter's start instead. Each clip says which: the moment picked, or why it was cut from the start.

They render in the background, and the page checks every 20 seconds. Each one shows its start time, its length and its state: preparing, rendering, ready or failed. A ready cut-down has a Download button. A short also has Post as a Short, which creates a YouTube Short draft in MOXA; as with every MOXA post, it goes out only after MOXA's checks and your approval, and the short then shows its post's state. If any cut-down failed, the button reads Make failed cut-downs again.

Cut-downs open once the finished video is ready. They are not made from a video too short to cut down, too large for the video renderer, or with no known length; the panel says which. Once the teaser is ready, a LinkedIn link post for a long episode carries the teaser.

Your website. Once the episode is published, its public page at /learn/<series>/episode-<week> plays the episode's YouTube video when it has one, and the finished video otherwise. It never plays the raw recording. The page also describes the video to search engines.

Your own site can list your episodes with its site key

If your website reads Nagent content with a site key, see connect your site, the same key reads your published episodes:

GET https://nagent.ai/api/cms/learning
GET https://nagent.ai/api/cms/learning?series=<series-slug>
GET https://nagent.ai/api/cms/learning/<series-slug>/<week>
Authorization: Bearer <your site key>

The list returns series, each with its slug, title, description, hostName and episodes. An episode carries its week, slug, title, excerpt, publishAt, videoUrl (the finished video), youtubeVideoId, durationSec, thumbnailUrl and chapters (each a title and a startSec). The single route returns one series and one episode.

With a valid key you get only your workspace's published episodes, never drafts and never another workspace's. A wrong, rotated or revoked key returns 401. As with the blog routes, a request with no Authorization header is not refused: it returns the public series of the site you called, so check the header is sent if you see episodes that are not yours.

What it costs and who can do it

• Shoot kit. About $0.20 per kit, as the Script step says.
• AI presenter. Each module's clip is checked against your workspace's budget before it is asked for and charged in credits when it is made. A refused check renders nothing. See credits and budgets.
• AI co-host. Live minutes are charged in credits. The budget is checked for a full 30-minute session before the co-host opens, and you are charged for the time the session ran.
• Cut-downs. Reading the finished video in the Knowledge Hub is charged to your workspace credits, like any video reading, and picking the best moments uses VIDYA's budget. Cutting and rendering the clips spend nothing more.
• Make the video, download and posting spend no credits of their own.

Making, rendering and posting need the content write permission. Anyone who can read content can download the finished video and its pack.

When you ask VIDYA in its team room instead of clicking, the same work runs as VIDYA's actions. Each one shows a card and waits for a person to approve it, whatever VIDYA's level (see autonomy and governance):

What the Beta does not do yet

• A digital twin as a long-form presenter.
• An AI tutor on the public episode pages.

Where to go next

• CREA and the content studio: the team VIDYA sits under.
• ALPHA personas and avatars: the presenter persona a series uses.
• MOXA: where the episode's posts wait for approval.
• Credits and budgets: what rendering and the co-host cost.