Building a multi-modal researcher with Gemini 2.5

The Gemini 2.5 series of models from Google has been at the top of leaderboards for many tasks with strong reasoning, coding, and multi-modal capabilities. In this video, we show how to take advantage of these capabilities, using LangGraph to build a multi-modal Gemini 2.5 researcher. Give it a topic to research and a related YouTube url (optional), and it will produce a short report as well as a custom 2-speaker podcast on the topic for you. It leverages Gemini 2.5's advanced capabilities for:

-  Integrated YouTube video processing
- Real-time Google Search integration
- Natural multi-speaker text-to-speech

Repo:
https://github.com/langchain-ai/multi-modal-researcher

Video notes:
https://mirror-feeling-d80.notion.site/Gemini-2-5-21e808527b1780c994fdde9349f448c3?source=copy_link