Google's new AI turns notes into podcasts!

PLUS: Adobe's Firefly video model

Author

Rowan Cheung
September 12, 2024

Welcome, AI enthusiasts.

Google's NotebookLM can now transform your documents into podcasts hosted by two AI agents with its new ‘Audio Overviews’ feature.

Turning your textbooks into audible SparkNotes has never been easier. Let’s get into it…

In today’s AI rundown:

  • Google turns your notes into podcasts

  • Adobe previews new AI video model

  • How to control AI video camera angles

  • Mistral releases multimodal Pixtral 12B

  • 6 new AI tools & 4 new AI jobs

  • More AI & tech news

Read time: 4 minutes

LATEST DEVELOPMENTS

GOOGLE LABS

Image source: Screenshot / The Rundown

The Rundown: Google just launched Audio Overviews, a new feature in NotebookLM that turns notes, PDFs, Google Docs, Slides, and more into AI-generated audio discussions between two virtual AI agents.

The details:

  • Audio Overview creates a "deep dive" conversation from uploaded sources, with AI hosts summarizing content and connecting topics across materials.

  • The tool works with various source types (documents, slides, charts, web URLs) using Gemini 1.5's multimodal capabilities.

  • To use Audio Overviews, open an existing notebook, navigate to the Notebook guide, and click on the “generate” button on the right-hand side.

  • We confirmed with Google Labs that NotebookLM can process up to 50 sources, each up to 500,000 words, allowing for a total of 25 million words to be considered when generating the audio.

Why it matters: Audio Overviews can be a game changer for people who are auditory learners. This feature works particularly well with academic papers, ebooks, textbooks, presentations — we tested it by turning yesterdays newsletter into an Audio Overview and were quite impressed!

TOGETHER WITH WEBAI

The Rundown: webAI offers a unique platform for local, customizable, and fully-owned AI models, empowering businesses to harness AI’s potential without compromising on security and flexibility.

With webAI, you can:

  • Develop and test AI solutions in-house, eliminating the need for external AI teams

  • Enhance security and reduce costs with local compute capabilities

  • Seamlessly deploy across various environments, from local to cloud and edge

Try webAI today to start building your custom AI.

ADOBE

Image source: Adobe

The Rundown: Adobe just previewed its Firefly AI Video Model, which includes tools to extend existing videos and create new clips from text or image prompts, coming before year end.

The details:

  • Three key features are coming: Text to Video, Image to Video, and Generative Extend.

  • Text to Video can create clips from prompts, with options for camera controls and reference images.

  • Image to Video transforms still images or illustrations into live-action clips.

  • Generative Extend, coming to Premiere Pro beta, can add footage to cover gaps or extend shots.

Why it matters: While OpenAI's (still unreleased) Sora focuses on generating videos from scratch, Adobe is aiming to create ‘a new era‘ for video editing itself. Soon, anyone will be able to change camera angles, extend scenes, and generate b-roll instantly with AI.

AI TRAINING

The Rundown: Luma Labs’ new ‘Camera Motion’ feature allows you to add professional-looking camera movements to your AI-generated videos with simple text commands.

Step-by-step:

  1. Log in into Luma Labs and access the main interface.

  2. Set up your scene by uploading an image or describing it in the text box.

  3. Type "camera" in your prompt to unlock the Camera Motion dropdown menu.

  4. Choose from options like Move, Pan, Push, Pull, or Orbit to add dynamic movement to your scene.

  5. Generate your video and iterate as needed to perfect the result.

Pro tip: Add starting and ending keyframes for even more control over your camera movements.

PRESENTED BY AGENT AI

The Rundown: Agent.ai is a new marketplace/network to find and connect with AI agents that can do a variety of tasks – such as automating company research, creating memes, and turning YouTube videos into LinkedIn posts.

With Agent.ai, you can:

  • Access a diverse marketplace of AI agents for many tasks

  • Gain hands-on experience with AI models and agent creation

  • Collaborate with a community of AI enthusiasts and innovators

Claim 100 free credits and start automating with AI agents.

MISTRAL

The Rundown: French AI startup Mistral has released Pixtral 12B, its first multimodal model capable of processing both images and text, available for free download under an Apache 2.0 license.

The details:

  • Pixtral 12B is a 12-billion-parameter model, about 24GB in size, built on Mistral's text model Nemo 12B.

  • Pixtral 12B is the startups first multimodal model, meaning it can process and answer questions about images, as well as text.

  • It's available for download via GitHub and Hugging Face under an Apache 2.0 license, allowing unrestricted use and fine-tuning.

  • Mistral plans to make Pixtral 12B available on their chatbot and API platforms, Le Chat and Le Plateforme, soon.

Why it matters: Despite being just over a year old, Mistral is rapidly emerging as Europe's “answer” to OpenAI. With a lean team of top researchers and a recent $645 million funding round, the company is aggressively pushing the boundaries of AI with powerful open models.

NEW TOOLS & JOBS

  • 🎥 Adobe Firefly Video - Turn text or image into video with the new Firefly Video Model

  • 🐼 Earkick - Free and anonymous self-care and AI chatbot

  • ⛰️ Playbook - Render 3D scenes with Stable Diffusion

  • 🎨 Verse - AI-powered visual blog, turn your idea into interactive webpages

  • 📄 Mintlify - Generates public documentation that converts users better

  • 🥑 Aicado - Easily integrate AI into websites with customizable models

QUICK HITS

OpenAI is reportedly in talks to raise $6.5 billion (plus another $5 billion credit line with banks) at a $150 billion valuation, significantly higher than its previous $86 billion valuation.

Meta is reportedly finalizing a new AI supercomputing cluster with over 100,000 Nvidia H100 chips to train its next Llama 4 language model.

Canada's Defence Minister announced the country is working on incorporating AI into its military by 2030, emphasizing for the tech to "improve the work of our military personnel, but not to replace it".

Kai-Fu Lee (former head of Google China), stated that Chinese AI models lag behind U.S. by 6 to 9 months, but predicts that Chinese AI apps will proliferate faster than in the U.S. by early next year.

Taylor Swift endorsed Kamala Harris for president, partly prompted by AI-generated deepfakes falsely showing her supporting Trump.

Oracle unveiled over 50 AI agents for its Fusion Cloud Applications Suite, designed to automate business processes across functions like capital management, supply chain, and customer experience.

THAT’S A WRAP

SPONSOR US

Get your product in front of over 650k+ AI enthusiasts

Our newsletter is read by thousands of tech executives, investors, engineers, managers, and business owners around the world. Get in touch today.

FEEDBACK

How would you rate today's newsletter?

Vote below to help us improve the newsletter for you.

Login or Subscribe to participate in polls.

If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.

Reply

or to participate.