Every New Google AI Update in One Video (NotebookLM, Gemini, and much more)

Key Concepts

• Agentic AI: Systems capable of autonomous reasoning, planning, and multi-step task execution.
• Multimodal Models: AI capable of processing and generating various media types (text, code, audio, video, images).
• Thinking Models: Advanced reasoning models (e.g., Gemini 3 Deepthink) designed for complex problem-solving and logic-heavy tasks.
• Vibe Coding: The process of building functional software or interactive tools using natural language prompts.
• Business DNA: A feature that extracts brand identity (fonts, colors, tone) from websites to automate marketing campaigns.

1. NotebookLM Updates

NotebookLM has introduced significant "agentic" features that automate content creation and presentation:

• Cinematic Video Overviews: Uses an agentic model to analyze source material, structure a narrative, and assign specific tasks to specialized models (e.g., Gemini 3 Pro for code-based animations, Nano Banana 2 for consistent visuals, and V3 for video generation).
• Factually Precise Visuals: Unlike standard video generators that hallucinate, this system uses code generation to create accurate diagrams (e.g., historical maps or computer science algorithms like Quicksort).
• Infographic Presets: New visual styles (Bento grid, clay, anime, etc.) allow for professional or creative data visualization.
• Slide Deck Editing: Users can now revise generated slide decks via natural language prompts, allowing for specific text removal or simplification.
• Context-Aware Generation: Users can generate infographics directly from the chat panel, ensuring the output is targeted to specific topics discussed in the conversation rather than the entire source library.

2. Music Generation Platforms

Google has integrated advanced audio capabilities into its ecosystem:

• Laria 3: A new music generation model available within Gemini (limited to 30-second clips) and the Producer AI platform.
• Producer AI (formerly Rift Fusion): A full-scale music platform that allows users to generate songs from prompts or lyrics and perform iterative edits (e.g., "make it darker," "add fiddle chaos").
• Functionality: It supports complex genre blending (e.g., Appalachian death metal) and allows for the creation of "Spaces" (interactive synths and drum machines).

3. Nano Banana 2 & Image Generation

• Accessibility: Free-tier users now receive 20 image generations per day (up from 2-3).
• Performance: Nano Banana 2 is significantly faster (10–15 seconds) and demonstrates superior text rendering and consistency compared to the Pro version.
• Thinking Mode: Essential for complex prompts (e.g., generating receipts or multi-object scenes). It prevents common errors like "extra fingers" or misspelled text.

4. Agentic Workflows (Manis & Google Workspace)

• Manis: An orchestration agent that manages multi-step research tasks. It can analyze YouTube comments, scour Reddit, and compile reports with B-roll.
• Skill Creator: A feature in Manis that records a user's workflow and packages it into a reusable "skill," allowing for automated, scheduled execution of complex tasks.
• Google Workspace Studio: Automates routine business tasks across Gmail, Sheets, and Drive, such as scheduling meetings or drafting documents based on email content.

5. Marketing & Browser Integration

• Pomelli: A Google Labs experiment that creates full marketing campaigns from a single product image. It includes an "edit" feature to fix physical inaccuracies (e.g., removing a lid from a bottle) and generates social media-ready assets.
• Chrome Sidebar: Now features Nano Banana 2 for direct image generation from web content and Auto Browse, which can navigate websites, filter search results (e.g., finding land for sale), and interact with forms.

6. Advanced Reasoning Models

• Gemini 3 Deepthink: Google’s flagship reasoning model for complex, high-level logic tasks, positioning it as a competitor to OpenAI’s o3 and Claude Opus.
• Gemini 3.1 Pro: A faster, highly capable model optimized for "vibe coding" and building interactive dashboards within Google AI Studio.

Synthesis

Google’s recent updates represent a shift from simple "chat-based" AI to agentic ecosystems. By integrating code-driven animation, automated marketing workflows, and advanced reasoning models, Google is moving toward a "digital employee" framework. The most significant takeaway is the move toward reusable workflows (via Manis skills or Workspace automation) and the prioritization of factual precision in visual generation through code-based engines rather than relying solely on probabilistic image models.