Do More Newsletter

This issue contains featured article "How Private Is the AI on Your Phone?" and exciting new product information about ElevenLabs Reader, Gamma, Napkin AI, Julius AI, and NotebookLM.

July 20, 2026

In partnership with

Keep up to date on the latest products, workflows, apps and models so that you can excel at your work. Curated by Duet.

Stay ahead with the most recent breakthroughs—here’s what’s new and making waves in AI-powered productivity:

ElevenLabs Reader is a mobile application that turns any text, article, or document into natural speech. Designed for everyday consumers, the app uses AI and advanced speech synthesis to read aloud in highly realistic voices. This allows users to listen to newsletter issues, books, or work documents while commuting, exercising, or multitasking. It is a fantastic tool for busy people who want to digest information on the go.

Gamma is a presentation and webpage builder powered by artificial intelligence. It enables businesses to quickly generate beautiful pitch decks, documents, and landing pages from simple text prompts. Users can refine layouts instantly without dealing with complex design software. This tool is perfect for creators who want professional results with minimal effort.

Napkin AI is an innovative visual creation tool that transforms text into relevant diagrams, charts, and illustrations. Users can paste their text or articles directly into the editor, and the artificial intelligence instantly recommends visual graphics that fit the context. This tool helps small businesses and creators turn dry documents into highly engaging visual presentations with ease.

Julius AI acts as a personal data analyst for business owners who do not have a technical background. By uploading spreadsheets, users can ask natural language questions to generate charts, uncover trends, and write code for data modeling. It simplifies complex business intelligence and provides actionable insights in seconds.

NotebookLM by Google is a personalized AI research assistant that helps users interact with their own documents. It acts as a virtual study partner, generating summaries, answering questions, and creating study guides directly from source materials. It is an excellent tool for self study, academic research, and deep project planning.

ElevenLabs Reader is changing how we consume written content by transforming text, using AI, into incredibly life like speech. This mobile application allows users to upload articles, books, research papers, or newsletters and listen to them narrated by highly expressive artificial intelligence voices. It acts as a personal narrator, turning long reading lists into a custom audio stream that fits perfectly into a busy daily routine.

The latest update introduces a powerful offline reading mode. This new feature allows users to download their favorite texts and synthetic voices directly to their device, enabling uninterrupted listening even when disconnected from the internet. Whether you are on a flight, commuting through subway tunnels, or taking a break in a remote area, you can continue listening to your critical documents without relying on cellular networks.

In addition to offline access, the update adds a custom voice sharing marketplace. Users can browse a curated selection of community created voices to find the exact tone, accent, and speed that suits their preferences. The app also features an intelligent smart bookmarking tool that syncs progress across devices, ensuring you never lose your spot when switching between your phone and tablet.

For small business owners, busy professionals, and consumers this tool maximizes productivity by recovering lost time. You can digest market research, competitor reports, and industry newsletters during your morning run or daily commute. By turning reading into a rich audio experience, ElevenLabs Reader helps professionals stay informed and productive without staring at screens all day.

How Private Is the AI on Your Phone?

Two years ago, Apple put ChatGPT inside the iPhone and called it a partnership. This month, Apple sued the company that makes ChatGPT for allegedly stealing its secrets.

In 2024, Apple wired OpenAI's chatbot into iOS and put it on stage like a proud new hire. On July 10 this year, Apple walked into a federal courthouse and accused OpenAI of poaching its engineers and walking out the door with confidential hardware designs — one former engineer allegedly failed to return his work laptop and later used a rare authentication bug to download dozens of confidential files, and job candidates were reportedly told to bring "actual parts" to their interviews. OpenAI says it has no interest in anyone else's trade secrets. The courts will sort out who's right.

But here's why it matters to you, and it has nothing to do with the lawsuit. While these two giants fight over who stole what, the assistant on your phone is quietly getting a new brain — one that's about to know more about you than any app you've ever installed. So it's a fair moment to ask a question most of us never do: when I talk to the AI on my phone, where does what I say actually go?

Back in February I wrote about what happens to the things you type into a chatbot. Think of this as the sequel nobody asked for — because the chatbot is about to be the phone itself.

I'll mostly use the iPhone here — it's where the shift is most visible this month, and where the courtroom drama is centered — but if you're on Android, the same questions apply, and I'll flag where Google does things differently.

What Actually Changed

This fall, the assistant on your phone starts getting rebuilt from the ground up. Apple's new Siri — and the "Apple Intelligence" behind it — isn't the old set-a-timer Siri. It's designed to reach into your actual life: your Messages, your Mail, your Photos, what's currently on your screen. Ask it "what was that address Mom texted me last week?" and the whole point is that it can go find it. (It arrives first as a beta, in English, on newer iPhones — but the direction is set.)

Google's rolling out the same kind of thing on its newer Android phones. This is the direction the entire industry is walking: an assistant that's genuinely useful because it can see your stuff.

And that's the trade. An assistant that knows nothing about you is safe and useless. An assistant that knows everything about you is powerful and — if you don't know how it works — a little unsettling. The good news is that "how it works" is simpler than the companies make it sound.

Your Phone Has Two Brains

Here's the one idea that makes all of this click.

When you ask the AI on your phone to do something, one of two things happens. Either your phone answers the question itself, right there in your pocket — or it mails the question off to a data center somewhere and waits for the answer to come back.

That's the whole distinction. On-device means it never left. Cloud means it did.

Picture a sharp personal assistant standing next to you. Simple stuff — "remind me at six," "find the photo of the receipt" — they just handle, quietly, without telling a soul. But hit them with something bigger, and they say "let me check with the home office," write your question down, and send it off. That note leaves the room. The question is no longer just between you two.

Modern phones do a surprising amount in that first bucket. The chip in a recent iPhone or Android flagship is powerful enough to run real AI without phoning home, which is faster and more private — nothing leaves. But the heavy lifting, the genuinely smart answers, still tends to go to the cloud. That's not a scandal. It's just physics: the biggest AI models are too large to live on a phone. It's worth knowing which questions stay in the room and which ones get mailed out.

The Promise Behind the Curtain

So what happens to the note once it leaves?

This is where Apple is making a specific, checkable promise, and it's worth understanding because it's the good part. When your phone sends a request to Apple's cloud, it's supposed to go to something called Private Cloud Compute — and the pitch is that the office reads your question, answers it, and then forgets it ever existed. Your data isn't stored. It isn't reviewed by an employee. It isn't kept to train tomorrow's model. And — this is the unusual bit — Apple lets outside security researchers inspect the system to check that the locks are real, rather than just asking you to take its word.

That's a genuinely higher bar than "trust us." Whether it holds up over time is exactly the kind of thing those independent security researchers exist to catch. But as promises go, "you can send someone to verify it" is a lot stronger than most of what you're offered online.

The Google Wrinkle

Now the twist that makes the lawsuit sting a little.

You'd assume the brains behind the new Siri are pure Apple. Mostly, they are — the AI models running on your phone and in Apple's cloud are Apple's own. But Google is tangled up in them in two ways worth knowing.

One: Google helped train them. Apple built its new models with heavy help from Google's Gemini, using Gemini as a kind of private tutor to sharpen Apple's models during development. Apple is insistent on the fine print — it says the finished models contain none of Gemini itself. Picture Gemini as the coach, not the player who takes the field.

Two — and this is the strange one: Apple's most powerful cloud model doesn't just learn from Google, it runs on Google — on Nvidia chips sitting inside Google's data centers. Apple says it stretched the same Private Cloud Compute promises over that rented hardware, so your data is supposed to ride Apple's privacy rails even while parked on Google's turf.

So: the models are Apple's, the biggest one runs on Google's machines under Apple's locks, and Google's AI helped raise the whole family. The myth worth killing is the scary one — no, Google's Gemini is not secretly running on your phone reading your texts. But step back and look at the board: Apple leans on Google while suing OpenAI, the partner it shipped ChatGPT with two years ago. Your most personal data now moves through a shifting web of companies that partner, compete, sue, and rent each other's technology all at once.

The safeguards may be excellent. The relationships are a soap opera. And that's the real reason to understand the settings on your own phone: not because any one company is the villain, but because you shouldn't have to track who's suing whom to know where your text messages are going.

The Work Problem Nobody Mentions

There's a wrinkle here that hits a specific group of people hard: what if your job flat-out forbids cloud AI?

Plenty of people work under exactly that rule — lawyers, healthcare staff, government contractors, anyone whose employer says company information may only touch AI that runs locally, never the cloud. For years that was easy enough to honor, because the AI was a separate app you chose to open.

That's the part that's changing. When the assistant is baked into the phone, some of what it does happens right there on the device — and some of it quietly gets sent out for a smarter answer. The tricky part is that you can't always tell which. So if that's your world, don't assume a feature is allowed just because it came built into the phone. The safe move is to check your employer's policy before you point any of it at work content.

Three Things to Check This Weekend

Enough theory. Here's what to do. It takes about five minutes, and you don't need to understand a word of the above.

1. See what the assistant is allowed to touch. On iPhone, open Settings → Apple Intelligence & Siri; on Android, open the Gemini app's settings. This is your control panel — where you decide whether the assistant is on at all, and which apps it can reach into. If you'd rather it never read a particular app, you can usually switch that off right here.

2. Make it ask before it hands your question to ChatGPT. On the iPhone, when Siri wants to send a hard question out to ChatGPT, there's a Confirm ChatGPT Requests switch (Settings → Apple Intelligence & Siri → ChatGPT). Turn it on and nothing goes to OpenAI without you tapping "okay" first — photos and files always ask. It's the one place you can actually watch the handoff happen and stop it.

3. Turn off "help improve" if you'd rather not. Buried in the same menus is an option to share your recordings, transcripts, or usage to "improve" the assistant. On Android it's typically on by default; on the iPhone it's usually something you opted into. Of the three, this one's nearly free: switching it off mostly just stops your voice and your questions from being kept to train the next version.

A couple of these come with a small trade-off — fence off an assistant and it gets a little less clever about you. That's the point, though: you're making the call out in the open instead of leaving it on autopilot. And if you want the receipts, the iPhone keeps them — Settings → Privacy & Security → Apple Intelligence Report shows you exactly which requests got sent to the cloud. It's a little technical, but it's proof the promise is checkable.

The Bottom Line

The companies will keep partnering and suing and swapping engines behind the curtain — that's their soap opera, and it'll run for years. You don't have to follow the plot. You just have to know one thing: the assistant in your pocket is either answering in the room or mailing your question out of it, and you're allowed to decide which.

The smartest AI on your phone isn't the one that knows the most about you. It's the one you actually understand. Take the five minutes. Then go ahead and let it find that address your mom texted you — on your terms, this time.

Partner Spotlight: AutoDoc by Duet Display

Capture every detail of your meetings without sacrificing your privacy. AutoDoc by Duet Display is a free meeting assistant that runs completely local on your device. Because it can operate even when disconnected from the internet, you can be sure that your private information never leaves your computer.

AutoDoc records both audio and video, and it intelligently syncs your notes directly to the timeline. This means you can not only read what happened in the meeting but also see exactly what was presented at that moment. AutoDoc is an open source project, but an installer is also available for download for those who just want to use it. Learn more at getAutoDoc.com.

What is stopping you from increasing your usage of AI?

Login or Subscribe to participate in polls.

That lead in your CRM? Gone.

Over 3.5 billion people open WhatsApp, Instagram, or Facebook Messenger every day. Your customers are already there, asking questions and comparing options.

Wati puts your business across WhatsApp, Instagram DM, Facebook Messenger, SMS, RCS, and web chat in one AI-powered inbox.

Automations instantly respond, route conversations, and keep every interaction tracked in one place.

Meet customers where they already are, before your competitor does.

Stay productive, stay curious—see you next week with more AI breakthroughs!