Every part of the pipeline is something you configure, not something you code.
Vuenia isn't a single script that makes one kind of show. It's a set of reusable building blocks — context sources, generation steps, verification rules, publish targets — that you arrange per Pipeline.
01
Pipelines and blueprints
A Pipeline is a named, reusable template: an ordered list of Steps, plus what feeds the first one. Start from a blueprint that matches what you're making, then customize it freely — add a step, remove one, swap a model, change who reviews what.
Basic
A simple, single-narrator audio or video on any topic.
Podcast
Multiple voices, a structured segment order, audio or video output.
Online course
A sequence of lessons — reading, video, and an auto-generated quiz. Coming soon.
Slide deck
The same content generation, exported as a deck instead of a video. Coming soon.
02
Context sources — bring your own data
Every Pipeline starts by gathering context. Configure any combination of live search, specific URLs, public or authenticated APIs and databases, or just a manual brief you type in. Define your own structure for how it's organized, or let Vuenia synthesize one automatically.
Search & web sources
Live research — news search, ticker-scoped queries, calendars, technical analysis.
Specific URLs
Point at a page or article and have it pulled in directly.
APIs and databases
Public or authenticated, for your own proprietary or licensed data.
A manual brief
Type or paste what this run should cover — no automated source required.
03
Generation steps
Each Step does one job — research, write, narrate, visualize, or assemble — and each is independently configurable: which model runs it, what prompt it follows, and what it's allowed to produce.
Script generation
A natural, on-brand script from your context, single-narrator or multi-voice.
Audio generation
Text-to-speech narration per character, with per-voice settings.
Slide generation
An LLM chooses and populates a fitting visual for every beat of the script.
Video assembly
Slides and narration stitched into one finished video, timed to the audio.
More, built in
Verification, built in
Attach fact-checking, date-checking, and custom rule agents to any step. Retry automatically with the specific violation fed back in, or escalate to a human after a set number of failed attempts.
Human review, wherever you want it
Pause after any step for approval, edit any output, and rerun a Pipeline from any point forward instead of regenerating everything from scratch.
Publishing
YouTube is available today, with generated title, description, and tags. Spotify and Instagram are coming soon.
Everything in settings
Voices, models, API credentials, publishing targets, verification rules — a new show or a new format is a configuration change, not an engineering request.