NewsPod AI

NewsPod AI is an example application of AuRA, our Autonomous Research Assistant (formerly IntellectCascade). It runs a news video podcast end to end. It monitors news sources, picks stories, researches them, writes a script, records narration, edits the episode with a lip-synced presenter, and uploads it to YouTube, all without a human in the loop.

How an episode gets made

1. Monitor. The agent watches news feeds for new stories in the themes a show is configured for.
2. Research. For each story it searches for further context so the script isn't limited to the original article, and it keeps track of the sites it used.
3. Write. An LLM turns the story and the research into a script, using prompt templates the agent adapts to the material.
4. Record. Text-to-speech voices the script, and a lip-sync engine animates a presenter avatar to match.
5. Edit and publish. The episode is assembled with music and title material, then uploaded with a generated title, description and tags, including a link to the research background.

Shows can be set to start compiling the news at a fixed time each day, and the uploader keeps its login session alive and reauthenticates on its own, so the channel runs unattended.

The series

The public playlist mixes several themed shows from the same pipeline: general AI news, an AI Launchday series on AI in education, politics, science and society, and an AI Stocks series on companies and markets. The same machinery also drives AI Storybook, which retells news for children. More episodes, the medical and SDG shows, and the research reports behind them are collected on AuRA's Outputs page.