- Everyday AI
- Posts
- Ep 652: NotebookLM: 8 New Updates for November that you should be using
Ep 652: NotebookLM: 8 New Updates for November that you should be using
Inside NotebookLM's latest updates, Anthropic announces its own infrastructure, OpenAI responds to NYT, Nano Banana lands on iPhone.
Everyday AI
November 12, 2025
š Subscribe Here | š£ Hire Us To Speak | š¤ Partner with Us | š¤ Grow with GenAI
In Partnership With Airia
Eliminate AI anxiety with Airia
Airia ia a unified, modelāagnostic enterprise platform to securely prototype, build, orchestrate, and govern AI agents at scale.
Move from anxiety to advantage with agent prototyping and orchestration in a safe environment, backed by enterpriseāgrade security, data governance, and automated testing that mitigates threats like data leakage and prompt injection.
Sup yāall š
Is there room for humans on the web in the future?
Weird question to ask, I know. But, it seems like tomorrowās web could be made by AI and FOR AI. And how can your brand still appeal to humans?
What are your thoughts?
Will a āpost-human webā (agents as primary consumers of content) become the norm?š³ļø Vote to see LIVE results š³ļø |
Weāre tackling this on tomorrowās Everyday AI show, so make sure to check back.
āļø
Jordan
Outsmart The Future
Today in Everyday AI
8 minute read
š Daily Podcast Episode: NotebookLM has dropped more than a dozen updates the past month. We cover the best. Find out more in todayās show and give it a watch/listen.
šµļøāāļø Fresh Finds: Elevenlabs new model drop, 97% cant spot AI music, Ex-OpenAI lead warns about AI Erotica and more. Read on for Fresh Finds.
š Byte Sized Daily AI News: Anthropic announces its own infrastructure, OpenAI responds to NYT, Nano Banana lands on iPhone and more. Read on for Byte Sized News.
š§ Learn & Leveraging AI: NotebookLM is much more than a personal AI assistant now. We break down how you can win back time today. Keep reading for that!
ā©ļø Donāt miss out: Did you miss our last newsletter? We covered: Inside Apple's $1 billion bailout from Google, ChatGPT getting group chats, Softbank sells it NVIDIA holdings, Microsoft downplays bubble fears. Check it here!
Ep 652: NotebookLM: 8 New Updates for November that you should be using
You probably missed these NotebookLM updates. š¤«
While AI World has been focused on bubbles popping and data centers, Google has kinda quietly rolled out a friggin cornucopia of updates to its Gemini-powered NotebookLM.
Real talk: NotebookLM is more than a personal productivity tool. It's now an essential AI secret weapon for growing teams.
Join us as we tell you how and highlight 8 new features you've gotta know. Let' put AI to work on Wednesdays!
Also on the pod today:
ā¢ Nano Banana visuals on steroids š
ā¢ Mobile quizzes and flashcards š±
ā¢ Google Sheets import, no sweat ā¬
Itāll be worth your 42 minutes:
Listen on our site:
Click to listen
Subscribe and listen on your favorite podcast platform
Listen on:
Hereās our favorite AI finds from across the web:
New AI Tool Spotlight ā Algebras Localizes your videos with AI, YouWare Transforms ideas into Projects, SuperApp builds native IOS apps with AI.
AI Music ā 97% couldnāt tell AI music from human tracks, and AI now makes up over a third of Deezer streams.
AI Agents ā AI āemployeesā hustleāand hallucinateāthrough a one-person startup experiment.
AI Chips ā Foxconnās Q3 beat, AI servers now outpacing iPhones, and it sees AI powering 2026 growth. An OpenAI-related announcement is coming next week.
AI Audio ā ElevenLabs released their updated text to speech model
AI Erotica ā Ex-OpenAI safety lead warns AI erotica claims are misleading.
Robotics and AI ā Russia's new humanoid AIdol falls seconds into its Moscow debut ā chaos on stage
LLMās and Hallucinations ā LLMs self-train on real documents, boosting reasoning without humans
AI Photo Editing ā Nano Banana brings advanced editing to Google Photos. Ask Photos expands to 100+ countries and 17 languages.
1. Anthropic to pour $50B into U.S. AI infrastructure šļø
Anthropic announced a $50 billion buildout of custom AI data centers in Texas and New York, with the first sites slated to go live in 2026 and partnerships led by Fluidstack.
The plan aims to secure U.S. compute capacity, add 800 permanent jobs and more than 2,000 construction roles, and accelerate the companyās frontier research as policymakers zero in on domestic infrastructure. This move counters OpenAIās massive long-term commitments with Nvidia, Broadcom, Oracle, and major clouds, raising fresh questions about power availability and whether the sectorās spend is outrunning the grid.
2. OpenAI pushes back on NYT demand for 20M private chats šµļøāāļø
In a timely escalation, OpenAI says The New York Times is asking a court to compel the handover of 20 million randomly sampled ChatGPT conversations from Dec. 2022 to Nov. 2024, a move the company calls an overreach tied to the Timesā ongoing lawsuit.
OpenAI argues the request risks user privacy and offers no real path to resolving the case, noting it previously beat back a broader bid for 1.4 billion chats and proposed narrower, privacy-preserving searches that were rejected. The company says the data would be de-identified, stored under legal hold, and viewed only by the Timesā outside counsel and consultants within strict protocols, while it accelerates plans like client-side encryption and automated safety systems.
3. Nano Banana lands as Google Photosā AI style engine, with iOS getting conversational edits š
Today, Google spotlighted its Nano Banana model that transforms simple prompts into polished photo styles, while the conversational āHelp me editā experience rolls out to iOS in the US with a redesigned editor.
Personalized edits powered by private face groups make iPhone tweaks more precise, and the āCreate with AIā carousel will start offering Nano Banana prompt inspirations on Android in the US and India next week. Coming soon, personalized templates will tailor style suggestions to each userās gallery. The new Ask button on iOS ties it together by letting you describe a look, get context on an image, and trigger edits right from the photo view.
4. Google unveils Private AI Compute for cloud-based, private Gemini processing āļø
Google today announced Private AI Compute, a new cloud platform that lets its most capable Gemini models run on sealed, hardware-protected infrastructure so users can get faster, more powerful AI without giving Google access to their personal data.
The system uses Google TPUs, Titanium Intelligence Enclaves, remote attestation, and encryption to keep sensitive inputs isolated and accessible only to the user, effectively applying on-device privacy guarantees to cloud processing.
5. AMD CEO backs Big Techās AI spending as a smart bet
Lisa Su told CNBC that heavy hyperscaler investment in AI infrastructure is not reckless but the right gamble, saying increased computing will speed innovation and that AMD expects 35% annual revenue growth over the next three to five years driven by āinsatiableā AI chip demand.
Her commentsācoming as megacaps disclosed more than $380 billion in AI spending and AMD shares jumped over 9%āsignal investor confidence in the profit potential of large-scale AI buildouts.
š¦¾How You Can Leverage:
NotebookLM dropped MAJOR updates in late October and early this month.
Radio silence. Aside from nonstop AI bubble talk.
Everyone's thinking of NotebookLM like "oh that's the thing that makes AI podcasts lolz" while completely missing that Google quietly shipped the only free AI chat with a full million token context window that's grounded by default.
Want to know the wildest part?
Companies paying thousands monthly for enterprise AI are getting worse context and more hallucinations than this free tool. And they have no idea because they think NotebookLM is a toy.
So on today's show, we broke down 3 of the most important recent NotebookLM updates that you should be using like yesterday.
The ones that save actual time and money. Let's get it.
1. Nano Banana Visuals Actually Understand Your Strategy š
We uploaded 50 pages about Apple paying Google $1 billion annually for AI.
Generated a video overview.
Nano Banana created a medieval scene. Apple on a cliff. Can't cross. Bridge made entirely of Google logos leading to an Apple castle.
Zero prompting for that metaphor.
The model read strategic dependency across those 50 pages and visualized the power dynamic better than any consultant deck we've seen. It's the number one image model on LMArena right now because it processes meaning instead of just matching patterns.
Your training materials with generic stock photos get 20% completion rates. Same content with strategically accurate visuals from Nano Banana hits 85% completion because people actually understand instead of just clicking through to finish.
Try This:
Pull your lowest-performing training doc right now and generate a NotebookLM video overview. Watch what visuals it creates without any custom prompting.
If they nail the strategic concepts your team struggles to grasp, you just found why mandatory training fails. Companies using this for onboarding cut time-to-productivity from six weeks to eight days because understanding actually happens instead of checkbox compliance.
2. 1M Token Context With Zero Hallucinations š„
ChatGPT maxes out at 128K tokens on paid plans.
Claude similar.
NotebookLM ships a full million tokens. Free. With grounding so it physically cannot hallucinate outside your uploaded sources.
That's not 8x better. That's a different category of tool.
We're talking full semester of college notes, 200 customer interview transcripts, complete product roadmap, and six months of board meeting minutes. All in one chat. Remembers everything. Never makes stuff up because everything's grounded in your sources.
Google also just increased multi-turn conversation capacity 6x last month, so extended back-and-forth stays coherent instead of degrading after message 12 like every other tool.
Your team currently spends 30-45 minutes hunting across docs for complex answers. With proper NotebookLM setup, that drops to 90 seconds with source citations that you can verify.
Try This:
Find the question your team Slacks about most where answering requires checking multiple long docs. Upload all those docs to one NotebookLM notebook.
Ask the question. Time the response. If you get accurate cited answer in under two minutes versus 45 minutes of document archaeology your team currently does, calculate weekly ask frequency.
Most teams realize they're burning 200+ hours monthly on searches one properly configured notebook eliminates completely.
3. Mobile App Plus Team Sharing Equals Compounding Intelligence š±
Flashcards and quizzes rolled to 100% of mobile users this week.
Combine that with persistent chat history, team sharing, and privacy controls that keep individual chats private even in shared notebooks.
Jordan preps podcast interviews at 2am when he can't sleep using mobile flashcards from research notebooks. No distractions in bed. Just drilling facts from sources already uploaded. Can't sleep anyway, productive instead of scrolling.
CEOs review board materials between meetings using flashcards from the real board packet uploaded days ago. Sales reps drill objection handling on the subway using data from their actual deals imported via Google Sheets. Zero desk time required.
Shared notebooks with proper privacy mean everyone builds on the same grounded foundation instead of recreating context individually every single time someone needs information.
Try This:
Create one shared notebook for a recurring team workflow where everyone wastes 20 minutes getting on the same page before actual work starts.
Add all relevant docs and set a clear goal for what the notebook should help with. Have three people test it for one week while tracking prep time.
If your 60-minute weekly status meeting drops to 35 minutes because nobody needs 25 minutes of context recap anymore, that's 100 hours annually reclaimed from one meeting fix. Scale that across every recurring workflow and the time savings get stupid fast.
Bonus Content
Not gonna lieā¦ it would be kinda foolish to not repost todayās LinkedIn livestream.
Weāve created about 30 āBonus Guidesā over the past 3 years and this is by far the most useful.
Reply