Modern Creator
Paul J Lipsky · YouTube

AI News: Fable Is Here; New Siri AI; NotebookLM's Biggest Update Ever; and Free New AI Tools!

A 28-minute weekly roundup that stress-tests Fable 5 on real work, unpacks Apple's biggest AI leap yet, and explains why NotebookLM just became a research agent.

Posted
today
Duration
Format
Tutorial
educational
Views
9.9K
435 likes
Part of the collectionThe Fable 5 PlaybookAll 45 Fable 5 breakdowns, synthesized into one page.
Read the playbook
Big Idea

The argument in one line.

Fable 5 writes in your voice better than any prior AI model, but for most automated agent workflows it is indistinguishable from Opus — making the 2x token cost hard to justify once the free window closes.

Who This Is For

Read if. Skip if.

READ IF YOU ARE…
  • You use Claude CoWork or Claude Pro and want to know whether Fable 5 is worth switching to before usage credits kick in.
  • You are an iPhone user wondering how Apple's new AI assistant compares to Gemini on Android.
  • You use NotebookLM for research and want to understand what the agentic upgrade actually enables.
  • You already pay for a Google AI Pro subscription and are considering adding the Fitbit Air for free health coaching.
SKIP IF…
  • You do not use Claude, Gemini, or Apple devices — none of the five covered tools will apply.
  • You want deep technical dives; this is explicitly a plain-language, no-code summary for everyday users.
TL;DR

The full version, fast.

Anthropic's Fable 5 is the most capable model they've publicly released, and it shows most clearly in tonal writing tasks — but in automated agent workflows it barely outperforms Opus, and at 2x the cost it's hard to justify once the free June window closes. Apple's WWDC26 AI push finally gives iPhone users what Pixel owners have had with Gemini: an assistant that reads your emails, messages, and calendar in context. NotebookLM's upgrade to a full research agent (with outside-web search from chat, new output formats, and cloud compute) is the week's most consequential change for knowledge workers, even though it is still rolling out to Ultra users only.

Free for members

Chat with this breakdown — free.

Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.

Create a free account →
Chapters

Where the time goes.

00:0000:53

01 · Cold open / intro

Teases the biggest AI updates ever; channel premise intro — AI news for regular people, no coding.

00:5301:00

02 · Subscribe CTA

Standard subscribe and bell notification pitch.

01:0002:08

03 · Fable 5 — what it is

Mythos backstory, guardrails explanation, benchmarks framed as secondary to real-world tests.

02:0805:22

04 · Fable 5 — real-world tests

Script completion in CoWork (impressive tonal match), website redesign (3 prompts to polished result), too-autonomous behavior observation.

05:2207:09

05 · Fable 5 — existing workflows

Health dashboard, email inbox adviser, CoWork research agent — minimal observable difference vs. Opus in structured tasks.

07:0908:15

06 · Fable 5 — availability and pricing

Free on Pro/Max/Team until June 22, then usage credits at 2x Opus cost; double-usage promo until July 5 softens the burn rate.

08:1508:56

07 · Fable 5 — verdict

Use it free for two weeks for creative/writing tasks; stick with Opus for ongoing workflows when it moves to paid credits.

08:5609:37

08 · Prompt/meme of the week

User connected WHOOP biometric data to work calendar via Fable to identify which coworker caused the most stress.

09:3711:22

09 · Sponsor — i10x

Multi-model AI workspace with access to OpenAI, Google, Anthropic, xAI, DeepSeek; image and video gen; super agent; $8-17.50/mo.

11:2213:38

10 · Apple — new Siri AI (WWDC26)

On-screen awareness, dedicated chat app with history, updated visual intelligence, iOS 27 photo cleanup demo (credit card removal), personal context across apps. Developer beta now, public beta soon, likely September release.

13:3815:48

11 · Google Gemini — third-party tools and price drop

Canva, Instacart, OpenTable added as third-party integrations. Google AI Plus drops from $7.99 to $4.99/mo with storage doubled to 400GB.

15:4816:35

12 · Gemini Live updates

Live now has personal intelligence memory (references prior chats); image generation added inside Live including from camera feed.

16:3518:54

13 · AI Gadget — Fitbit Air

$99 no-subscription fitness tracker. Google AI health coaching free with Gemini Pro subscription. Generates custom workouts, sends proactive messages, aligns feedback to stated fitness goals.

18:5423:56

14 · NotebookLM — full research agent

Agentic capabilities: web search from chat with source import prompt, new output formats (PDF/DOCX/XLS/PPT/charts), visible thinking, Gemini 3.5 + Antigravity, 100+ cloud software skills. Rolling out to Ultra subscribers first.

23:5624:02

15 · Subscribe CTA — NotebookLM video teaser

Promises a full NotebookLM test video once Ultra access arrives.

24:0225:41

16 · Google AI Edge Eloquent

Free on-device dictation cleanup tool. Local Gemma 2B or 12B. Removes filler words and self-corrections. Live demo shows Thursday/Friday slip auto-removed.

25:4128:06

17 · Dream Beans

Google Labs app that reads Gmail, Calendar, and Gemini chat history overnight and generates personalized curated story cards each morning. Watercolor image style.

28:0628:51

18 · AI School community waitlist CTA

Closing pitch for upcoming Claude CoWork community with systems, health dashboard, email automation, competitor research templates.

Atomic Insights

Lines worth screenshotting.

  • Fable 5 can match a creator's tone of voice well enough that readers cannot tell where the human stopped writing and the AI started.
  • Fable 5 costs twice the tokens of Opus, so the current double-usage promo until July 5 is effectively the only window to use it at parity pricing.
  • When Fable is used in existing automated workflows, the real-world difference versus Opus is nearly zero — the intelligence gain shows on tonal tasks, not structured agent tasks.
  • Apple's new Siri can read your emails, messages, calendar, and reminders simultaneously when answering a question — that cross-app context is what has been missing from every prior Siri.
  • Google AI Plus dropped from $7.99 to $4.99/month and doubled storage from 200GB to 400GB — the cheapest credible AI subscription on the market.
  • NotebookLM can now search the open web from inside the chat and ask if you want to import sources, which was previously impossible from the chat interface.
  • NotebookLM's agentic capabilities are grounded in user-provided sources, which may reduce hallucination compared to using bare Gemini for the same research task.
  • The Fitbit Air ($99, no subscription) includes Google AI health coaching for free if you already pay for Gemini Pro — effectively a free add-on for existing subscribers.
  • Google AI Edge Eloquent does on-device dictation cleanup using local Gemma 2B/12B — no audio sent to the cloud, and it removes filler words and self-corrections in real time.
  • Dream Beans reads your Gmail, Calendar, and Gemini chat history overnight and surfaces personalized stories each morning — the Google version of a proactive AI briefing app.
Takeaway

What each AI release actually changes — and what it does not.

WHAT TO LEARN

The week's biggest releases each have a real use case and a hard limit, and knowing both prevents chasing updates that will not move your workflow.

  • Fable 5 is genuinely better than Opus for tonal writing tasks, producing output close enough to a creator's voice that readers cannot identify the seam — but this advantage disappears for structured agent workflows where Opus performs nearly identically.
  • Fable 5 costs twice the tokens of Opus; the current double-usage promo through July 5 is the only window where the economics are neutral, making right now the best time to evaluate it for your highest-value writing tasks.
  • When Fable acts as an agent, it can be over-eager — interpreting a clarifying question as a directive to start implementing, which is a meaningful risk in automated workflows you do not want changed without explicit instruction.
  • Apple's new Siri is a qualitative leap because it reads emails, messages, calendar, and reminders simultaneously in context — the cross-app awareness that Android users have had with Gemini on Pixel and iPhone users have never had before.
  • NotebookLM's shift to a full research agent matters most because it keeps AI responses grounded in sources you control, which reduces hallucination risk compared to running the same multi-step research task in a general-purpose chat interface.
  • Google AI Plus at $4.99/month with 400GB storage is now the lowest-cost credible AI subscription available; pairing it with a $99 Fitbit Air gives Gemini Pro subscribers a full AI health coaching system with no additional monthly fee.
  • On-device dictation cleanup (AI Edge Eloquent) removes filler words and self-corrections before text is inserted, with no audio sent to the cloud — useful for privacy-sensitive contexts or anyone who wants cleaner dictated notes without a subscription.
Glossary

Terms worth knowing.

Fable 5
Anthropic's newest and most capable publicly released Claude model, derived from the more powerful but unreleased Mythos model, with additional safety guardrails.
Mythos
An internal Anthropic model considered too dangerous to release publicly; Fable 5 is a guardrailed version of it.
Claude CoWork
Anthropic's agentic platform where users can run automated workflows, agents, and projects beyond simple chat.
Usage credits
A pay-per-use billing layer on top of a subscription plan; once Fable 5 moves to usage credits in late June, subscribers pay extra per token beyond their base plan.
NotebookLM
Google's AI research tool that grounds answers in documents you upload as sources; the new version adds agentic web search and multi-format output generation.
Agentic capabilities
The ability of an AI tool to take multi-step autonomous actions — like searching the web, creating files, and running analysis — without requiring a separate prompt for each step.
Google AI Edge Eloquent
A free Google tool that transcribes speech on-device and uses a local Gemma language model to clean up filler words and false starts before inserting text.
Dream Beans
A Google Labs app that reads your Gmail, Calendar, and Gemini conversation history overnight and generates personalized curated story cards each morning.
Antigravity
A Google model mentioned alongside Gemini 3.5 as powering NotebookLM's new thinking and reasoning capabilities.
Gemma
Google's family of small open-weight language models designed to run locally on a device; used in AI Edge Eloquent for on-device text cleanup.
Resources

Things they pointed at.

09:37producti10x
11:22productApple Intelligence / new Siri (WWDC26)
13:38toolGemini third-party integrations (Canva, Instacart, OpenTable)
16:35productFitbit Air
24:02toolGoogle AI Edge Eloquent
25:41toolDream Beans
Quotables

Lines you could clip.

02:55
I think if I showed you this entire script, you wouldn't be able to know where I stopped writing and where Fable started.
Concrete, verifiable claim — strongest Fable endorsement in the videoTikTok hook↗ Tweet quote
08:28
If I have to pay extra to use a model, I find myself just not using it. And I think that's going to be the case here.
Relatable honest pullback after positive review — the tension is the hookIG reel cold open↗ Tweet quote
23:16
Would you use NotebookLM for this? This feels like a task for Gemini.
Contrarian question in the middle of positive coverage — cuts through hypenewsletter pull-quote↗ Tweet quote
The Script

Word for word.

Read-along

Don't just watch it. Burn it in.

See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.

analogystory
00:00This week, we have seen some of the biggest AI updates ever to both Claude and, believe it or not, Apple's assistant. Google also released some understated but very important updates to both Gemini and Notebook LM, and they released some free new tools.
00:15One that is very impressive that you guys are gonna love, and the other one that's probably their weirdest release so far. This was a huge week.
00:24We got a lot to cover. If you don't know who I am, my name is Paul. Here in this channel, I talk about the AI news that regular people actually care about.
00:33We don't talk about coding. We talk about how AI can help you in your everyday life or with your work. Now I typically do these AI news updates every single week for you.
00:43I took a couple of weeks off because of some travel, but I'm back now ready to get you caught up on all the AI news, and I'll be returning back to my regularly scheduled once a week updates. So if you wanna stay informed about the latest AI news that actually matters, make sure to subscribe and turn on the bell notification.
01:00Let's start with the biggest news first. Anthropic dropped a new model called Fable five. Now a couple of months ago, they announced that they have created a model called Mythos, which is so powerful they said they cannot release it because in the wrong hands, it could be very dangerous.
01:17So Fable is basically a version of Mythos but with more guardrails. Anthropic has said that this is their most powerful model yet that they have ever publicly released at least, and the benchmarks certainly support that.
01:32But if you've been following my channel for a while, you know that I don't care about benchmarks or marketing hype. I wanna see how it performs on real world tasks. So first, like I mentioned before, there are a lot of guardrails with Fable and you might feel those guardrails with certain queries.
01:46If you try to do something Fable doesn't like, you'll get a message saying that you can't ask Fable that and to use a different model instead. There have been a lot of people on X complaining about this. Personally, I have not had this happen to me once just having it do general knowledgeable tasks for me.
02:02So let me show you what I've been able to do with this and my opinion on how good Fable is. So the other day, I was working on this script pretty much when Fable came out. And I had to run to the gym or I wanted to run to the gym.
02:18I didn't finish writing the script. I didn't even get the test fable out. So what I asked fable to do was to look at the script and to just finish it for me.
02:27I wanted to look at the first half to understand my tone of voice. I said don't make it AI slap slop. I don't want any cringe language.
02:34I don't want it to sound like AI. You know, make it in my tone of voice. And then I just left and went to the gym.
02:41Now, normally, I don't have AI write my scripts for me because it always comes out really bad. Just like this very predictable AI language that I hate that doesn't sound good. But Fable did a really good job.
02:58The best I've seen any AI do when it comes to writing. And it's very usable. It's not perfect, but it's very usable.
03:06So this line right here, for instance, it says, this is the part that made Cowork finally click for me, so I want to spell it out. Like, is classic AI language. That's kind of sloppy.
03:17But the rest of it is really excellent, and I was super impressed with it. Like, I think if I showed you this entire script, you wouldn't be able to know where I stopped writing and where Fable started.
03:28That's the first time I've seen that. It's really incredible. Not perfect, like I said, but the closest version I've ever seen to my tone of voice.
03:37I also asked it to redo my website for me. So this was my original website, and I always thought it looked fine.
03:45There was nothing wrong with it. It looked pretty good. One thing you'll notice here is that you can't see any images.
03:51Like, the image of my face is normally here, and the image of the YouTube videos are here as well. Those don't render when I'm on my, uh, computer like this.
04:01They only render when I actually have them on my actual website. So just kinda keep that in mind. But when I had Fable actually go through it, and I just said, hey.
04:10Just basically make this better. Update it and make it better. That was the only instructions I really gave it.
04:16And I think it made it a lot better. So first of all, it somehow figured out how to get the images, like, permanently in there so that I don't have to have them as a separate file, which is really nice. And it just kinda subtly improved it.
04:29So we have, like, this more, uh, moving dynamic image here with this border, These things popping out, this hand waving.
04:38It just fits more of the style that I was going for and feels a lot nicer than the original one that we had going, which again was fine, just not as good as this new version. Had to go back and forth with it a little bit because I wanted some personal changes, like I wanted to be permanently in dark mode. But also, it did get some stuff wrong, like some of the logos were incorrect and there was a spacing issue.
05:03But overall, it was really good, and I was able to get it looking like this after just about three prompts. So I was very impressed with this.
05:12I think it looks a lot nicer than the original one. Now there have been a few strange quirks that I've noticed with Fable since I've been using it. So first, it almost feels too autonomous.
05:24A few weeks ago, I had Opus set up for me this health dashboard to help me with my health goals. I'm trying to gain some muscle. So it's helping me track that, and it even sends me text messages throughout the day to make sure I'm staying on track.
05:36Now this morning, I got a message from my agent that I didn't quite understand, so I asked it for clarification. And instead of clarifying, it just started implementing changes to the system, which is not what I wanted it to do.
05:51It just assumed that it knew why I was asking instead of just answering my question. And I've noticed that in general, it just seems almost a little bit too eager.
06:01It also seems incredibly intelligent, sometimes too intelligent, and I can't always understand what it's talking about. There's also plenty of workflows that I have running.
06:11So you see I have a daily AI brief, an email inbox adviser, nightly cleanup, and all my my health ones going.
06:20And I noticed that when I switched over to Fable, I didn't really notice any difference. So in conjunction with this dashboard, I have my health coach. I was hoping that it would be able to give me more intelligent insights into my workouts and my progress, but I didn't really notice any difference between Opus and Fable.
06:39I also have co work doing research for me. It creates this research report for me every morning that has the latest AI news and different topics like that that I may be able to cover on my YouTube channel. It's always been really good at doing the research part, but not very good at the analysis or the understanding of what would make a good video topic, and that continues to be true with Fable as well.
07:00There's a certain understanding there that no AI I've worked with has been able to figure out yet, and that's true for Fable as well. Let's also talk about availability.
07:09Fable is right now included with your pro max and team subscription through January 22.
07:18After that, you'll be able to use it with usage credit. So it will cost additional on top of your subscription.
07:26At some point though, it will come back to subscription plans. We just don't know when that's going to happen. Fable also costs twice the price of Opus.
07:35So you're gonna burn through your usage twice as fast. However, right now, uh, Anthropic has a deal going on where you get twice the usage inside of Cowork until July 5.
07:48So effectively, you should be able to use Fable just as much as you were using Opus. So we have a great opportunity right now to try it out for the next couple of weeks until we switch over to usage credits.
08:00So the real question is, does Fable even matter? How is it going to impact the work that you do with Claude or Claude CoWork, and should you be using it at all? I personally really like it, and I'm going to continue to use it for the next two weeks while it's part of my subscription.
08:15Once it moves out of the subscription though and I have to start paying extra to use it, I think I'll only use it for my most important tasks. But generally speaking, anytime I have to pay extra to use a model, I find myself just not using it. And I think that's going to be the case here, especially because Opus is still really good.
08:34And for most day to day tasks, you're not gonna notice a huge difference. So if you wanna have some fun testing out Fable, go ahead for the next two weeks. But otherwise, if you have your workflow set up and running and you're happy with them and the results you're getting, then stick with Opus for now and maybe explore Fable when it permanently is added as part of our subscriptions.
08:56Okay. Let's move on now. Every week during my weekly news update, I have a few segments that I include as part of the update.
09:04And two of them are the prompt of the week and the meme of the week. The meme of the week is just something funny that I found in the world of AI that I wanted to share with you. So this week, I'm going to combine both of them into one.
09:16I saw this post by this guy, uh, p a n k a j, if you wanna check them out. So what he did was he hooked his WHOOP up to his work calendar to find which coworker gives him the most stress. And using Fable, he was able to pull all the heart rate data out of WHOOP and match it up.
09:35So I think that I thought that was pretty funny. So I'll have this post down below if you wanna check it out. He has a blog where I think he shares how he set this up.
09:43Now like I said, the second big announcement I wanna cover this week was from Apple. But before we get to that, let's hear from today's sponsor I 10 x. I 10 x is an all in one platform that gives you access to all the major AI models in one workspace.
09:59So when you start a new chat, instead of being limited to the models from one company like OpenAI, you also get access to models from Google, x AI, Anthropic, and DeepSeek. You can also turn on things like web search, memory, and knowledge bases.
10:16If you wanna create images, they have image generation using other models like NanoBanata two, Image two from OpenAI, and Flux.
10:26And you can even generate videos using leading models. So we have Klingon here, v o 3.1, and XAI.
10:33They just added a new super agent, which you can check out right here. With this, you can set up these different templates so you can have your agents research for you, track competitors, set up whole marketing campaigns.
10:45And as you see, it connects to all the third party apps that you already use and love. I 10 x starts at just $8 a month if you go with annual billing or $10 a month if you pay month to month. If you get the unlimited plan though, it's just $17.50 a month, and that gives you unlimited usage of basic models and advanced models and 75 credits for things like image and video generation.
11:09If you wanna sign up for it, I'll have a link in the description down below along with a coupon code to get an additional 15% off. And thank you I ten x for sponsoring this video. So the second big AI update from the past week was from Apple.
11:22They have finally updated the AI capabilities on all their devices, and we have a brand new assistant, which they're calling this.
11:32I'm just going to call it the assistant from now on though, so I don't accidentally trigger your devices. So this assistant will be integrated throughout your iPhone, your MacBook, your iPad, your watch, your Apple Vision Pro, and it's unlocking some really nice looking capabilities.
11:50So first, it will have on screen awareness. So you can invoke it from anywhere and ask you questions like, show me photos from when I went to Spain. It'll just show you those photos.
12:00You'll be able to take action through multiple apps like messages, music, reminders, and more based on what you're doing in that moment while still being able to tap into broad world knowledge. There's also going to be a dedicated assistant app that you can see all of your past chats, and this is kind of like a ChatGPT clone in some ways.
12:24So you'll be able to have a running list of all the chats that you've had with it. There'll be an updated visual intelligence where it's just where you can aim your camera at something and have it give you information like the nutrition inside the food that you're eating.
12:38And, of course, it's gonna help you with things like writing. They've also updated their image playground, which was a tool for generating images, and it looks much better than the existing one, which quite honestly is pretty much trash. And there are new capabilities inside the photo app.
12:53So if you take a photo and you wanna reframe it, you can actually move around the image to reframe it, and cleanup tool has gotten a big improvement. This person had a chance to test it, so they held this credit card in front of their face. They asked the new iOS 27 to erase the credit card, and look.
13:12It's done a fantastic job. The big difference here with what they've implemented is the fact that Apple is going to have so much context about you.
13:21You'll be able to read your emails, your messages, your reminders, your calendar, and pull all of that information in when it's answering your questions.
13:31It feels very similar to what Gemini is on Pixel devices, but now it's finally for iOS. So this is currently in developer beta, which means that developers have access to it. And there will be a public beta soon, and then it will be released to the public, I think, around September.
13:49So I will be joining the public beta and probably will make a video on this once that is available to me and I've been able to test it out. Let's move on now to what Google has been up to with Gemini. If you come into the connected app section of your Gemini settings, you'll see down here that it now works with more third party tools like Canva, Instacart, and OpenTable.
14:12So with Canva, for instance, you'll be able to search, summarize, and design digital assets with Canva inside of Gemini. Now in and of itself, this is not huge news.
14:21This is only three new apps. But the broader implications of this are significant because for a long time, Gemini only worked with Google products like Gmail.
14:33They have said that they are opening up support for more third party tools. So this is good news. We are finally seeing the first steps of this.
14:42And personally, I think Gemini needs to add more third party tools if they wanna be competitive. Another huge win is that the price of Google AI plus has just dropped from $7.99 a month to just $4.99 a month.
14:59And the only change to the plan besides the price is actually that they are doubling the storage from 200 gigabytes to 400 gigabytes. It's kind of mind blowing that you get that much for just $5 a month now. So personally, I've never used a plus plan myself because I've always had pro and ultra plans, but I do plan to buy it on one of my accounts and compare it against the cheapest ChatGBT plan, uh, which is like $8 a month or something like that.
15:31So I'll be comparing both of those plans in a video. So if you're interested in that, make sure to subscribe. That is coming soon.
15:37There have also been some significant updates to Gemini Live. I'm not sure when this first one actually came out, but I just noticed it the other day. When you are using Live, it now has access to personal intelligence.
15:49So it can reference prior chats and knowledge about you when answering questions, which makes it so much more helpful. The other update is that inside of Live, you can ask it to create images for you, and it can do that now, including when you're sharing your camera.
16:05So you can point your camera at something, ask it to transform what you're looking at, and it can do that.
16:11It's really neat. Next, we have our AI gadget of the week. This is a segment where I talk about some new AI gadget that I've been trying out and showing you if it's worth you buying or not.
16:23So this week, we are talking about this device right here. It's called the Fit bit Air. This is made by Google.
16:30They bought Fitbit a few years ago. So this is a very thin band that is kind of like a whoop, if you're familiar with that.
16:40So it's fitness tracker, and it only costs $99, and there's no subscription fee with it.
16:46You get access to all your basic data with just the device. You don't have to pay anything month to month.
16:54However, you can upgrade to get Google Health coaching. That does cost money, but Google Health coaching is included with a Google AI Pro and Up subscription.
17:06So if you're already paying for Google AI Pro, like Gemini Pro, you get access to the health coach. I've been using this for a few weeks now, and I am very impressed with it. So it is, of course, a great fitness tracker.
17:19There's nothing super surprising there. It tracks your steps, your calories, your workouts, your sleep, things like that. But what I really like about it is the AI coach.
17:27So for instance, earlier today, I needed to do a workout. My gym is closed today for some reason. So I just open up the coach and told it, hey.
17:36I have to work out at home. Here's the equipment I have available for me. Now it not only put together a plan, but it actually put together a workout for me that I was able to follow.
17:46So you can see it has the warm up, the entire circuit, and the cool down. And if you click on start workout, it'll actually walk you through this workout.
17:54After your workouts are done, you get a message kinda summarizing it and telling you how well you did and how well it aligns and brings you closer to your goals. Because the first time that you use it, it will sort of interview you and ask you questions. So it knows that I'm really into ninja, and that's the main thing that I'm training for.
18:12So everything sort of evolves around that. So it says, hey. It's good that today you didn't do ninja.
18:17You did, you know, like, nice, uh, HIIT workout, and it's you know, this was your heart rate during it. And this is probably gonna help you a lot with, like, your grip strength, especially carrying I did I had to do a bucket carry for me, like a heavy bucket carry.
18:29So that's gonna help with my grip strength for ninja. So it's really cool. It kinda brings all that together into these proactive messages after your workouts.
18:37It'll also message you at the start of the day and the end of the day to let you know, like, what you have going on for the day and kind of a recap of how everything went. So I am very impressed with it. I really love it.
18:47And I especially love that after I buy the device, it's completely free because I'm already paying for a Gemini subscription. Next, we need to talk about what has been going on with Notebook LM. So first, two minor updates.
19:01Anytime you make something inside of Studio over here on the right, if you open it up, you'll notice on the top left, you can now view the prompt and sources. So if you're wondering what sources were used to make this artifact, now you can see exactly which ones were used, which is pretty nice to have that and also be able to easily access the prompt.
19:22The second minor update is that sources, if you have any sources that are inside of Google Drive, will automatically sync.
19:31This wasn't the case in the past if you added, let's say, a Google Doc or a Google Sheet. Then in the past, it would kinda take a snapshot of that. So it if you made any updates to the Google Doc, those wouldn't be reflected in the notebook.
19:45But now they are reflected. Reflected. It It automatically automatically syncs.
19:48Syncs. The big update though is that NotebookLM is now a full research agent.
19:55Here's what they had to say about this announcement. They said this is a new powerful NotebookLM, and they called it a massive upgrade that delivers agentic capabilities in chat, more advanced reasoning, and a suite of new output formats.
20:10We'll talk about that in a second. So this should be able to tackle complex multistep research problems a lot easier than in the past.
20:18So if we scroll through this video that they made, we can see some of that in action. So it initially looks exactly the same to me. I'm not gonna lie.
20:26But as you begin to use it, here for instance, they're asking the chat to search for things that aren't already in the notebook. Then the notebook goes out, does the research, and finds outside sources and asks if you wanna import them as a source, which you can do directly inside the chat.
20:45That's something you cannot do inside of NotebookLM right now. To add outside sources, you can either add them manually or use the research tool over here on the left. You can't do it from the chat.
20:56NotebookLM will also be able to create new artifacts. So here it's creating a PNG, and then here it's creating an Excel file, and you'll also be able to see it's thinking.
21:07So if we look at it thinking right here, you can see it actually go through the thinking process. We see this with other AI chats. Now that's coming to Notebook LM.
21:16In a follow-up post, Josh Woodward, who's a VP at Google, has said that some of those output formats will include PDFs, Docx, Excel files, PowerPoints, charts, etcetera.
21:27And all of those, you'll be able to make just by asking the chat to make them for you. So that part of it, I get that's really exciting. What I don't quite fully grasp yet are the agentic capabilities and exactly what that brings to Notebook LM.
21:42They have gone on to say that this new improved Notebook LM will be powered by Gemini 3.5 and surprisingly antigravity.
21:52So that's what enables that thinking process. Plus, each notebook will have a secure cloud computer including 100 plus curated software skills, unlocking deeper research and more complex analysis, which sounds great, but what exactly does that mean?
22:09So here in this example that they gave, the person asked, can you compare our gym sales to what we spent on ads? And they added in as a source a few documents, I guess that shows their sales and their ad spend, and they're having Notebook LM compare the two of them.
22:27So that's really cool and all, but it really makes me question, would you use NotebookLM for this? This feels like a task for Gemini, and I'm not sure it makes sense to do this inside of NotebookLM and what the benefit of doing it inside of NotebookLM is.
22:45The only thing I can think of is because the way that NotebookLM is set up, it is grounded in the sources that you provide. So perhaps setting it up inside of NotebookLM and running it inside NotebookLM will mean that it will hallucinate less than if you did it inside of Gemini.
23:04Until I get my hands on this and can really test it, I'm really not sure what the use case is going to be. So as for availability, they have said that right now it is only rolling out for ultra users, but very soon or at least down the line it will come out to other users, other plans in the future.
23:24Now I have two Google AI Ultra plans, and I do not yet have access on either account to these updates. So I don't really know where it is or what's happening with it.
23:36I don't know anyone who has access to it yet. So I'm eagerly awaiting it. And, of course, once I do get access to it, I will thoroughly test it and make an entire video about it because I do think this is a big deal.
23:48We just have to understand, like, the nuance of it and how we're gonna use this and when we should use this over other tools like Gemini. So if you don't wanna miss out on that video when I create it, make sure to subscribe to the channel and turn on the bell notification. And finally, I wanna talk about two new tools that were released by Google this past week.
24:08This first one, you're gonna absolutely love, and both these are free by the way. This one is called Google AI Edge Eloquent. So this is an AI dictation tool, But what makes it unique is that it not only transcribes your voice, but it cleans it up, and everything is done on device.
24:23It's not sent to the cloud. So it uses a local transcription model and Gamma or Gemma, was it four b or 12 b, which are two local models.
24:34Yeah. I'm using Gemma for two b. You can use 12 b.
24:39These are used for the cleanup. The 12 b model works better, but it's a little bit too slow for my taste. And I found the two b model still works really well.
24:48So let me show you it in action. So let's say you have an email that you're writing. I can invoke it by pressing this button.
24:56You see it down here on the bottom, and I can start talking to it. So let me restart that.
25:01Hey, Susan. Thanks for getting back to me. Can you meet with me on Thursday I mean, on Friday?
25:10And there you go. You see it has the text, and it got rid of my ums, and it got rid of the Thursday part and just left the Friday part.
25:20So it works pretty well. And, this is free. It's definitely not perfect.
25:24Like, I hate that it excluded the beginning part where I said, hey, Susan. I also don't like that the animations, it sticks down here way too long. There's supposed to be a setting where this goes away and it kinda works inconsistently.
25:37So it's not perfect, but it's pretty nice for being free. And the second free new tool that they've released is kind of a weird one.
25:45It's called Dream Beans. So the idea is this provides proactive, personalized collections of stories each day covering things that matter most to you.
25:55So what that means is that in the middle of the night, this app will go through your Gmail, your calendar, things you've chatted about with Gemini inside the Gemini app, and create for you these stories that are curated for you that might teach you something new or give you suggestions on things to do.
26:15So I got access to that. This is my dream beans for today. And it says, try making a thick super food acai bowl at home.
26:24Not a 100% sure why it surfaced this, but it is certainly very relevant for me. Maybe that's because when I was in San Diego recently, I kept asking it, where can I find the best acai bowls? Or maybe it's because inside the Google Health app, I've been logging a lot of smoothie bowls that I've been eating, uh, after my workouts.
26:43If you click on the story, it'll open it up, and you can read it. So here it's telling me to make a superfood acai bowl, and there's a little recipe for it, things I can put inside of it.
26:53And there are always these links where you can explore more about this topic or even ask Dream Beans about it if you need some clarification or follow-up things. Every morning, you get a few stories like this. So here's my acai bowl story.
27:08Then here's one oh, it kinda moves fast on my phone or my computer. Here's one where it's telling me to check out these smart glasses that I've used before and how I might be able to use them for early morning focus and tracking board for the latest Notion updates.
27:24I guess I've talked with it about how I've used Notion in the past. So it's pretty neat. It's definitely interesting.
27:32Right? With these labs tools. So this is part of Google Labs.
27:37These often start out as, like, these strange experiments, and they give it their own app. And a lot of times, it gets incorporated into the the bigger app.
27:47So I can see this being part of Gemini in the future. I don't know if it really needs to be its own app, but I think for, like, right now as we're testing out, it makes sense. And I guess it's it's pretty cool.
27:57I also like the images that it creates and, like, this watercolor style. It's definitely a lot better than doom scrolling in the morning. I could just kinda go through these curated fun stories.
28:06So that's all the AI news from the past week. This was a big update. The last thing I have for you is a reminder that I'm going to be opening up my AI school community very soon.
28:17This is where I'll share all my exact Claude CoWork systems that I use to run my business, generate my health dashboard I showed you before and coach, to answer emails for me, and to do my competitor research. Of course, you'll also be able to connect with other community members and learn from them as well. To get on the wait list for that community, make sure to click the link in the description down below.
28:38Joining the waitlist doesn't require you to actually join the community. It just shows that you might be interested, and then I'll email you later with more details. Again, the link for that is down below.
28:48Thanks so much for watching, and I'll see you in the next video. Bye for now.
The Hook

The bait, then the rug-pull.

A week this packed usually means scattered noise — but Paul Lipsky opens with a specific claim: Claude, Apple's assistant, and Google all moved in ways that regular people will actually feel. The title leads with Fable but the payload is broader.

Frameworks

Named ideas worth stealing.

02:08concept

Fable evaluation framework

  1. Tonal writing tasks — significant improvement
  2. Structured agent workflows — minimal difference
  3. Availability window — free until June 22
  4. Cost after window — 2x Opus, burns credits twice as fast

A four-axis way to decide whether to switch to Fable 5 from Opus: type of task, observable difference, current window, and marginal cost.

Steal forEvaluating any new premium AI model release against your existing workflow
CTA Breakdown

How they asked for the click.

VERBAL ASK
28:06product
I'm going to be opening up my AI school community very soon... to get on the wait list, click the link in the description.

Low-pressure waitlist ask — emphasizes no commitment required. Preceded by a genuine content summary of what members get (CoWork systems, health dashboard, email automation).

MENTIONED ON CAMERA
FROM THE DESCRIPTION
OTHER LINKSAlso linked in the description.
Storyboard

Visual structure at a glance.

open
hookopen00:00
Fable reveal
valueFable reveal01:00
CoWork demo
valueCoWork demo02:08
pricing details
valuepricing details07:09
verdict
valueverdict08:15
Apple Siri
valueApple Siri11:22
Gemini integrations
valueGemini integrations13:38
Fitbit Air
valueFitbit Air16:35
NotebookLM agent
valueNotebookLM agent18:54
AI Edge Eloquent
valueAI Edge Eloquent24:02
Dream Beans
valueDream Beans25:41
CTA
ctaCTA28:06
Frame Gallery

Visual moments.

Watch next

More from this channel + related breakdowns.