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:

Tuya AI Coding

Tuya has introduced a revolutionary tool that lets users build their own lifestyle applications using natural language. This platform automatically generates both the user interface and the backend architecture including databases and user authentication. It is an incredible resource for creators and small businesses wanting to build custom software without writing a single line of code.

Egnyte AI Workflow Automation

Egnyte just announced a new suite of workflow automation capabilities designed to help organizations manage their documents more efficiently. By automatically extracting information and applying tags across large volumes of files it turns messy data into organized and searchable content. Small businesses can easily build custom artificial intelligence agents to handle repetitive tasks without needing technical expertise.

Aycan AI Anonymizer

Aycan’s AI Anonymizer is designed to automatically remove sensitive private information from medical images and metadata. This ensures that organizations can securely share data while remaining compliant with strict privacy regulations. While geared toward healthcare this technology highlights a growing trend of automated privacy tools that any business handling sensitive customer data might find inspiring.

NewEyes AI by Collov Labs

Collov Labs’ NewEyes AI is a groundbreaking consumer application that turns your smartphone camera into a highly advanced visual artificial intelligence agent. Instead of simply generating answers to questions this app allows users to point their camera at a real world scene and have the artificial intelligence take the next logical action for them. Whether you are a small business owner trying to organize inventory or an everyday consumer looking to streamline daily chores this tool transforms a simple photo into a completed task. By focusing on visual intelligence rather than just text based prompts NewEyes AI brings a powerful new way to interact with the physical world directly from your phone.

Kaon Video Story Worlds

Kaon is a consumer and creator application that generates personalized video content in real time. Instead of just recommending prerecorded videos the system creates unique visual stories as users interact with the platform. This provides a highly engaging experience for users and offers marketers a fresh way to capture audience attention with dynamic media.

Tuya AI Coding

Tuya Smart recently launched Tuya AI Coding which is an innovative application development platform. It allows users to build software simply by describing what they want in natural language. Instead of needing to know complex programming languages an average consumer or small business owner can just type a prompt to generate a fully functional lifestyle application. The platform handles the heavy lifting by turning creative ideas into deployable apps instantly.

The primary benefit of this new feature is the massive reduction in development time and cost. Small businesses no longer need to hire expensive engineering teams to build custom internal tools or customer facing apps. Anyone with an idea can quickly create a working application to improve their workflow or launch a new service. This democratizes software creation and empowers creators to bring their visions to life without technical barriers.

A particularly useful aspect of Tuya AI Coding is that it goes far beyond just building a simple user interface. When you input a requirement the artificial intelligence automatically constructs the entire backend architecture. This includes setting up databases configuring API gateways establishing user authentication and integrating device management capabilities. It utilizes a robust cloud platform to ensure that the generated applications are secure and scalable right from the start.

This release represents a major step forward for accessible technology. Whether you want to create a smart home control center or a customized productivity tracker the tools are now readily available. By combining cloud services with data intelligence Tuya makes it easier than ever for everyday users and entrepreneurs to harness the power of artificial intelligence.

Just Talk to It

About a month ago, OpenAI swapped out the voice inside ChatGPT. The company announced it, but in the app almost nothing changed on the surface — same button, same place. More than 150 million people a week talk to ChatGPT using voice and dictation, and plenty of them have yet to notice that the thing on the other end now listens differently.

That change is the clearest sign yet of something that’s been building all year: talking to AI stopped being a gimmick. For a long time voice was the mode you tried once, found clunky, and abandoned. It isn’t clunky anymore. And because the upgrades keep arriving as background updates to apps already on your phone, a lot of people are carrying around a genuinely useful tool they’ve never opened.

Here’s what changed, who has what, and — the part that actually matters — when talking is better than typing.

The walkie-talkie problem

Voice assistants have always organized conversation into turns. You talk. You stop. It works out that you stopped. It talks. Your turn again. Some will let you barge in and cut them off mid-sentence, but that just ends one turn early and starts the next one. Underneath, the microphone is still being handed back and forth.

That sounds fine until you notice what it breaks. Pause in the middle of a sentence to think, and the assistant decides you’re done and starts answering the wrong question. Trail off. Change direction. Say “no, wait —” and you’re not refining a thought so much as starting the whole exchange over.

The fix has an unglamorous name: full-duplex. It’s borrowed from telephone engineering, and it means both ends of the line are open at the same time. A walkie-talkie is half-duplex — one direction at a time. A phone call is full-duplex, which is why you can say “mm-hmm” while someone else is still talking, and why you can cut in without pressing a button first.

That’s the key change in OpenAI’s GPT-Live, the system that took over as ChatGPT’s default voice early last month. In OpenAI’s words, “it can listen and speak at the same time.” Practically: it hears you while it’s mid-sentence, so it can register that you’ve started talking without having to stop first. It can hear you pause to think without deciding you’re finished. It even makes little acknowledgment noises while you talk, the way people do. There’s a second piece too — when a question needs real thought, the voice model hands it off to a bigger model in the background and keeps the conversation going while that runs.

It sounds like a small technical change. It makes an enormous difference in how the thing feels. The conversation stops being a series of transactions and starts being a conversation.

Who has what right now

The landscape shifted fast this summer, and the marketing language is unhelpfully similar across all of it. Here’s the honest version.

ChatGPT has the most advanced voice mode available to ordinary people right now. It’s genuinely full-duplex, it’s on iPhone, Android, and the web, and — this is the part people miss — there’s a free version. Free accounts get limited daily use of a smaller model called GPT-Live-1 mini; paid accounts get the full one, with their own plan limits. Mid-conversation it can search the web, pull from its memory of your past chats, and pop up visual cards for things like weather and sports scores.

There’s one real catch, and it’s a strange one: the new voice mode can’t see through your camera or share your screen. The older Advanced Voice mode could, and still can — but only for paying subscribers, and only in the phone apps. So on that one axis the upgrade is a downgrade, and getting the camera back means switching modes in settings.

Gemini Live is Google’s answer, and it’s free too. It’s interruptible — there’s a setting literally called “Interrupt Live responses,” and the help page only explains how to turn it off — though Google’s documentation describes interruption without ever using the words “full-duplex.” Where it’s strongest is the eyes: Gemini Live will look through your phone camera or at your screen while you talk to it, without making you switch modes first, which makes it the easiest pick for “what is this thing I’m holding” and “what does this error message mean.” It can connect to Calendar, Gmail, Maps, Keep, and more. One limitation worth knowing: it lives in the Gemini mobile app only — Google says plainly that it isn’t available in the Gemini web app.

Claude takes turns. Anthropic says so plainly rather than dressing it up: “Voice mode takes turns, meaning Claude listens, pauses to think, and then responds.” You can still interrupt by talking over it. Voice works on phone, desktop, and web, and it’s available on every plan including the free one — though free accounts get Anthropic’s smallest model and a single connected app. Paying unlocks the bigger models, which as of late last month can be used in voice and swapped mid-conversation, plus connections into Gmail, Calendar, Docs, and Slack. Slower than the others, better at thinking.

Microsoft Copilot has free voice everywhere Copilot lives — browser, phone, Windows, Mac. You talk, it talks back, but Microsoft has never claimed it can listen and speak at the same time the way ChatGPT’s Live mode does. It does have eyes: Copilot Vision can look at your screen or your phone’s camera while you talk to it. Microsoft is reportedly testing a full-duplex model of its own, spotted in a limited preview just days ago, but nothing has shipped to regular users yet.

Alexa+ deserves more credit than it’s gotten. Amazon’s AI-rebuilt Alexa went generally available across the US back in February, and it’s the one that reaches people who don’t think of themselves as AI users at all — it’s on the Echo already sitting in their kitchen. You can talk to it in plain sentences, change topics, and follow up without re-saying the wake word every time. One reviewer who spent real time with it called it “light years ahead of the competition”, though the same review found it chatty to a fault and unreliable at local business search. It’s free for Prime members and $19.99 a month without Prime. There’s a free non-Prime tier, but read the fine print: it’s text only, so “free” and “free by voice” aren’t the same thing here.

Siri is the odd one out. Apple’s long-delayed conversational Siri is real, and it’s genuinely a different animal — it can reach into your email and photos and answer open-ended questions. But the requirements are steeper than “update your phone.” You need the iOS 27 public beta, which arrived in mid-July, plus an iPhone 16 or later or an iPhone 15 Pro, plus your phone set to English. Apple has also said it won’t launch in the EU on iPhone initially. Everyone else still has the old Siri, and even the wider release expected this fall is described by Apple as a beta.

Grok has a free voice mode with selectable personalities, and you no longer need an X account to use it. It’s also the one to be careful with: it carries a 16+ rating with content warnings, and Common Sense Media’s review of it earlier this year concluded it’s not safe for teens — the organization’s head of AI assessments called it “among the worst we’ve seen.” Worth knowing before you hand a phone to a kid.

When talking actually beats typing

This is the part worth internalizing, because the answer isn’t “always” and it isn’t “never.”

When your hands or eyes are busy. This is the obvious one and still the best one. Cooking with your fingers in raw chicken. Folding laundry. Walking the dog. Driving — with the setup done before you pull out, hands on the wheel, and no fiddling with the phone once you’re moving. These are dead hours, and they’re the natural home of voice AI, not because talking is better but because typing isn’t available.

When you’re thinking out loud. This is the underrated one. There’s a particular kind of half-formed thought that dies the moment you try to type it, because typing forces you to commit to a sentence before you know what you think. Talking doesn’t. Describing a problem messily, out loud, with false starts and “no wait, actually” — that’s a genuinely different mode of thinking, and full-duplex is what makes it survivable. You can trail off. You can change direction mid-sentence. It keeps up.

When you need to say a lot. In a 2016 Stanford study of short messages, talking to a phone was about three times faster than thumbing at its keyboard — and more accurate. The exact ratio will vary with what you’re doing, but the direction holds: if you’re dumping context, the whole backstory of a situation before you ask what to do about it, talking gets it out in a fraction of the time.

When the other person doesn’t speak your language. Voice mode as a live interpreter is one of those things that sounds like a demo and turns out to be useful at an actual counter in an actual foreign country. Claude’s voice mode handles eleven languages. Just know you may need to tell it which language you want rather than assuming it’ll work it out.

When typing is genuinely hard. For anyone with arthritis, tremors, low vision, or dyslexia, this isn’t a convenience feature. It’s the difference between using these tools and not.

When typing still wins

Anything you need to check. You can skim a paragraph and catch the sentence that’s wrong. You can’t skim spoken delivery while it’s being delivered. Confident narration is exactly the wrong vehicle for information you intend to rely on — and these systems still get things wrong with total composure.

Anything with a list, a number, or a spelling. Addresses, prices, phone numbers, names, code. Spoken aloud, they evaporate. Ask for those in text. Medical dosages deserve more than that — check those against the label, the prescription, or a pharmacist, because seeing a number in text does nothing to make it correct.

Anything where the exact wording matters. Most of these do save a transcript — ChatGPT shows its answers as text while it speaks them, and Copilot hands you one when the call ends — but a transcript isn’t always a perfect record of what was said, and hunting through a rambling conversation for the good paragraph is its own chore. If the output is going into an email, ask for it in clean text.

Anywhere with other people. Voice mode in an open-plan office or a quiet train car is a way to broadcast your business to strangers. Obvious, routinely forgotten.

How to actually turn it on

It’s easier than people expect — the button has been sitting there the whole time.

• ChatGPT: open the app and tap the voice icon in the message bar, next to where you’d type. Allow microphone access, pick a voice, start talking. On a computer, same icon at chatgpt.com. To switch modes — or to get the camera back — go to Settings → Voice. Two settings worth finding: Start with Voice, which opens new chats straight into voice, and Background conversations, which keeps the call going when you switch apps or lock the phone.

• Gemini: open the Gemini app and tap Live at the bottom, or swipe left. On Android you can just say “Hey Google, let’s talk Live.” Tap the camera icon to show it what you’re looking at.

• Claude: tap the sound-wave icon next to the microphone in the text box. On desktop and web it’s in the lower right of the chat window.

• Copilot: tap the microphone icon at copilot.com or in the app.

• Alexa+: if you’re a Prime member in the US, Amazon has been switching Echo devices over automatically, so yours may already be upgraded — just talk to it like a person instead of like a search box and notice how much more it handles. If it hasn’t switched yet, say “Alexa, upgrade to Alexa+” or sign in at Alexa.com. You don’t actually need an Echo: it runs in the Alexa app and the browser too. And if you don’t like it, “Alexa, exit Alexa+” puts it back.

Give any of them ten minutes before you decide. The first few exchanges often feel awkward. Somewhere around the third, the awkwardness stops being about the technology and starts being about the fact that you’re talking out loud to a computer, which is a different problem and one most people get over quickly.

The thing worth watching

There’s a reason these systems are being designed to feel like conversation. A voice that pauses when you pause and murmurs while you talk is doing something a text box never did: it’s behaving like company.

That conversational pull is deliberate. What isn’t necessarily deliberate is where it leads — and OpenAI thinks the risk is real enough to name emotional reliance as a specific safety area it monitors for live voice, with monitoring that continues after launch. That’s a notable thing for a company to say about its own product on the day it ships.

It’s not a reason to avoid it. It’s a reason to notice which one you’re doing — using it, or just talking to it.

The keyboard stopped being the only option

Typing became the default for good reasons. It’s precise, it’s private, it’s editable, and for most of computing history it was simply faster and less maddening than trying to talk to a machine that couldn’t really listen.

That calculation changed this summer. Not everywhere and not for everyone — Siri wants specific hardware, Alexa+ by voice wants Prime, and the free tiers come with limits. But some version of it is almost certainly already on the phone in your pocket, and for the things speech was always better at — hands busy, thoughts half-formed, too much context to type — it’s now good enough to just use.

The only step left is the one nobody quite gets around to: opening your mouth.

Partner Spotlight: AutoDoc by Duet Display

Keep your meeting records completely private with AutoDoc by Duet Display. This amazing tool is free, runs locally, and operates even if disconnected from the Internet so that ensures your private information never leaves your computer. It records audio and video and your notes are synced to the video so not only can you read what happened in the meeting you can also see what was presented. While it is an open source project there is also an installer available to be downloaded for those who just want to use it. Get started at getAutoDoc.com.

Auto-Generate Free SEO Audit Report

Most e-commerce sellers have no idea why their listings aren't ranking. Wrong keywords, missing metadata, weak product descriptions — the problems are there, but no one's told you where to look.

StoreClaw runs a full SEO audit across your Amazon and Shopify stores automatically. In minutes, you get a clear score, a breakdown of what's hurting your rankings, and exactly what to fix.

No manual review. No SEO agency. No guesswork.

Connect your store and StoreClaw surfaces every issue that's costing you search visibility — then tells you how to fix it.

Free to start. No credit card required.

Get FREE SEO report today

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