Do More Newsletter

Sponsored by

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:

Meta AI Mac App

Meta launched its new desktop application for Mac users, giving creators and small business owners direct access to AI assistance right from their macOS toolbar. The app connects directly to Facebook, Instagram, Meta Ads, and Google Workspace, helping users audit ad performance, analyze post engagement, and draft documents without switching browser tabs. Small businesses can monitor account metrics in real time while using system wide voice dictation to compose messages anywhere on their computer.

BeautyPlus AI Influencer Generator

Pixocial released a virtual persona creation tool on its BeautyPlus Web platform, designed for creators and small marketing teams seeking consistent digital branding. The application solves character drift in generative media by maintaining exact facial features, hair, and style across entire batches of photos and social videos. Marketers can generate ready to post visual content from prebuilt lifestyle personas or custom characters without arranging physical photo shoots.

Kimi Work for Mac

Kimi Work by Moonshot AI for macOS, is an intelligent desktop assistant built to automate complex workflows and local file management for solopreneurs. The tool allows users to set plain text objectives through Goal Mode, enabling the assistant to search local documents, synthesize online research, and organize files across apps. It provides a single home for multi step research and administrative tasks without requiring manual data transfers.

Friendly AI Assistant

Friendly Technologies’ Friendly AI Assistant, a consumer application that removes technical friction from managing connected home networks and smart devices. Users can speak or text simple requests like "fix my lagging television" or "why is WiFi slow," and the AI analyzes real time network data to execute automated fixes. The tool bridges conversational AI with home router management to keep everyday devices running smoothly.

OtterlyAI Agent Analytics

OtterlyAI’s Agent Analytics give site owners and small business marketers clear visibility into the AI crawlers visiting their websites. The software analyzes server logs to reveal requests from web fetchers like ChatGPT, Claude, Perplexity, and Google, tracking which pages are scanned to power AI search answers. Marketers can now track invisible bot traffic, measure AI search return on investment, and optimize content for generative engines.

Meta AI Brings Screen Aware Intelligence to the Mac Desktop

Meta recently released its standalone Meta AI desktop app for Mac, bringing screen aware intelligence and business tool integrations straight to macOS users. Built natively for Apple silicon, the application gives small business owners and creators a dedicated workspace companion that stays accessible right on the desktop. Users can trigger the assistant instantly with keyboard shortcuts to ask questions, draft communications, or manage ongoing tasks without opening a web browser.

The defining capability of this release is the new live window sharing feature. By granting screen permissions, users can select any active open window on their desktop and let Meta AI read visible text or analyze captured visuals in context. Rather than copying and pasting lengthy documents or uploading attachments manually, you can display a spreadsheet, a design layout, or a draft email and immediately ask the assistant for tailored analysis, structural recommendations, or summaries.

For small business operators and content creators, the feature integrates directly with professional Facebook and Instagram accounts, Meta Ads, and Google Workspace. This means the window reader can look at live campaign metrics or social post analytics on your screen and pair that visual data with backend account performance. You can instantly evaluate audience reach, calculate return on ad spend, or draft follow up social posts tailored to top performing formats.

To make daily execution even faster, the app includes cross platform dictation and floating composer controls. Pressing the quick invoke shortcut displays a compact prompt box over any active workspace, allowing you to dictate prose directly into Mail, notes, or project managers. With low memory overhead and deep system integration, this new tool transforms passive desktop viewing into an active workflow engine.

Who Taught Your Kid to Use AI?

School is starting across the country — already underway in a lot of places — and this year plenty of districts are handing out something new: a written policy on artificial intelligence. What students may use it for. What they may not. What counts as cheating now.

It’s worth reading. It’s also, on its own, not close to enough — because the policy covers the part of this that’s easy, and stops well short of the part that isn’t.

The number worth sitting with

Common Sense Media surveyed 1,204 kids ages 9 to 17 and published the results in June. Eighty-six percent say they use or interact with AI.

Read that number carefully, because it’s broader than it sounds. It counts the AI summary at the top of a Google search, writing helpers like Grammarly, photo editors, and school tools — not just chatbots. The chatbot figure is lower and still striking: two-thirds of kids have used one. Nearly a quarter of kids use some form of AI every day.

That’s the headline, and it’s the least interesting finding in the study.

Here’s the one that matters: more than four in ten kids say no parent or guardian has ever talked with them about AI safety. Not “talked about it recently.” Ever.

So the technology is nearly everywhere. The guidance is not. Kids can pick up the mechanics by trial and error, the way most of us learned to use search engines — typing things in and seeing what came back. That worked out fine for search. Search didn’t answer in the first person.

The rules came first

Schools haven’t been idle here. Seventy-three percent of kids say their school or a teacher has told them what they should and shouldn’t use AI for in schoolwork — which, given how fast this showed up, is a real accomplishment.

Safety is the newer piece: 56% say their school has discussed how to use AI safely, and only about half have been told how to judge whether what AI tells them is accurate. That ordering makes sense. A school has to answer the integrity question first, because that’s the one that lands on its desk in September. What counts as your own work, what gets a zero, what a teacher does with a suspicious essay — those are institutional problems, and they have institutional answers.

The rest of it is harder to put in a handbook. It was never really the school’s to carry alone.

The legislation follows the same logic. FutureEd is tracking 77 bills across 27 states this session that deal specifically with AI in classroom instruction, and ten have been enacted:

• Oklahoma’s Responsible Technology in Schools Act requires every public school district to adopt a written AI policy before the 2027–28 school year, bars AI from being the primary basis for grading, discipline, placement, promotion, or retention decisions, and preserves a parent’s right to opt their child out of student-facing AI tools without academic penalty.

• Alabama will require public school students, starting with the class of 2032, to earn a computer science credit to graduate — either an approved course or an approved course with computer science embedded in it — under a state definition that now explicitly includes artificial intelligence. It takes effect October 1.

• Utah updated its existing grade 7 or 8 digital skills course, directing the state board to have standards in place by 2027–28 covering AI literacy, cybersecurity, and digital privacy.

• New York has companion bills that would bar most classroom AI use before ninth grade, carving out diagnostic uses and specified interventions for students with disabilities. Both are still sitting in committee.

That’s a tracker built to cover teaching and learning, so it’s no surprise the bills concentrate on coursework, governance, privacy, and accountability. They’re the right things to legislate. They also have nothing to say about what happens at 11 p.m. on a personal phone, and nothing in that category was ever going to.

One more marker of where this is heading: in 2029, the OECD adds a new domain to PISA — the international exam comparing 15-year-olds across countries — called Media and Artificial Intelligence Literacy. It’ll be the first time this gets measured across borders.

The homework question is the easy one

Most parents’ instinct is to police the assignment. Did you use AI on this? That question is a dead end. It’s binary, it’s adversarial, and it invites exactly one answer.

The better question is “What did you use it for?” — because the honest answers sort into two very different piles. It explained the thing my teacher went too fast on is a kid using a tutor. It wrote my conclusion paragraph is a kid using a ghostwriter. Same tool, same session, completely different outcome, and no policy document can tell them apart from the outside.

The toolmakers know it, and they’ve started building for it. ChatGPT’s Study Mode walks a student through a problem with guiding questions instead of handing over the answer, and Google’s equivalent in Gemini is called Guided Learning. Both are worth switching on and showing your kid once.

Then, last week, OpenAI began rolling out ChatGPT for Teens. Buried in it is a feature called homework reminders — the system tries to recognize when a teenager appears to be shortcutting an assignment and nudges them toward Study Mode instead. There’s also Study Hours, where a parent or the teen can pick windows when eligible new chats simply start in Study Mode.

That’s a genuinely useful piece of engineering. It’s also an admission. You don’t build a feature that detects a kid trying to skip the work unless a great many kids are trying to skip the work.

The hard one is what else they’re asking it

Back to the survey, because this is the part that doesn’t make the orientation packet. These next figures describe kids who use AI, not all kids — which, at 86%, is nearly all of them anyway.

Fifty-seven percent have asked AI for information or advice about their health or their body. Thirty-seven percent have used it to discuss their feelings or personal problems. Among that group, one in four say they sometimes feel the AI understands them better than most people do.

Among kids who’ve used AI chatbots specifically, 17% say one has shown or said something that wasn’t OK for someone their age. Only a third of them told a trusted adult.

Now set that beside what the companies themselves say. Google’s guidance page for parents managing a child’s Gemini access states it plainly: Gemini “should not be treated as a friend or used to discuss personal problems or secrets.” Google also notes, in the same breath, that its content filters “try to limit inappropriate content from appearing, but they’re not perfect.”

So the advice exists, in writing, from the company that built the thing. It’s on a help page almost no child will ever open.

Whether an AI can be a real companion to anyone is a genuinely open question, and a contested one. But that argument is somewhat beside the point here. What matters is the scale: how ordinary this has become among 9- to 17-year-olds, and how much of it can happen entirely outside adult view.

What parents can control now

Both companies have built real machinery for this, and it’s easy to miss.

ChatGPT is rolling out automatic placement into the teen experience when account-provided age, verified age, or age prediction indicates a user is under 18. Availability is expanding over the coming weeks, so it may not have reached your kid’s account yet. Parents who link an account can set Quiet Hours, schedule Study Hours, manage selected settings, and receive notifications in limited high-risk situations. Notably, parents do not get to read their teen’s conversations. OpenAI’s under-18 rules also go past blocking explicit content: the model isn’t supposed to use romantic language, encourage emotional dependence, or imply that it has feelings or consciousness.

Worth knowing if your kid is on the younger end of that survey: ChatGPT isn’t meant for children under 13 at all, and OpenAI requires parental consent for 13- to 18-year-olds.

Gemini requires a parent to switch it on through Family Link before an under-13 child can use it, and applies added filters for minors — including filters meant to stop it from claiming to be a person or having human emotions. Google publishes a Family Guide to AI that’s short enough to actually read.

All of it still depends on the platform identifying your kid’s age correctly. OpenAI combines account-provided age, verification, and prediction; Google relies on account information and Family Link supervision. A made-up birth year may still slip through — but it doesn’t automatically defeat these systems the way it once would have.

Three things worth doing now

Ask the school three questions. Which AI tools are approved for my kid’s classes? If we opt out, will my child face any academic penalty? What happens to the data my kid types in? If your state passed something this session, some of the answers may already be written down.

Ask your kid one question, and not the accusatory one. What do you use it for? Ask it like you’re curious, because you should be. The answer tells you which pile your kid’s use falls into, and it opens a door that “did you cheat” slams shut.

Link a parental-controls account. Not to read their chats — you can’t, and that’s by design — but because linking is what turns on Quiet Hours, Study Hours, and the safety notifications. If you’ve never done it, that’s the single highest-value ten minutes available to you here.

The label on the bottle

The most telling line in last week’s announcement isn’t a feature. It’s that OpenAI now says, in writing, that its under-18 rules direct ChatGPT not to imply it has feelings or consciousness.

That rule exists because enough kids needed it. The companies can put a label on the bottle. They can filter the worst of it, nudge a kid back toward the work, and remind everyone that the thing talking isn’t a person.

What they can’t do is be the adult who already asked.

Partner Spotlight: AutoDoc by Duet Display

AutoDoc by Duet Display is a free productivity tool designed to capture and organize your meetings effortlessly. AutoDoc runs completely local on your machine and operates even when disconnected from the Internet, guaranteeing that your private information never leaves your computer. The app records audio and video while syncing your meeting notes directly to the video timeline, allowing you to read what happened and visually review what was presented at any moment. AutoDoc is an open source project, but an easy installer download is available for anyone who wants to start using it immediately. Find more information or try it now at www.getAutodoc.com.

Smarter CRM. Less Busywork.

Disconnected data and tools make it harder to understand your customers. HubSpot's Agentic Customer Platform brings your data, teams, and tech stack together with AI built in to help your business work faster and create more personalized customer experiences.

Why HubSpot and what's new

• Use AI powered tools to take action faster

• Unify your data, teams, and tech stack in one place

• Create one shared view of customer data

• Connect teams around the same customer context

• Bring your business tools into one place

Connect more of your business in one place and give every team a smarter way to work. Get set up quickly and start checking off your hardest tasks.

Get started free

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