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:

OpenAI Expands Codex to Mobile for On-the-Go Development

OpenAI rolled out its Codex coding assistant directly into the ChatGPT mobile app this week, giving developers and technically curious users the ability to manage software projects from their phones. The move turns smartphones into portable AI workstations where users can review code suggestions, approve changes, ask questions about projects, and even trigger development tasks remotely. While the initial support focuses on macOS systems, the broader significance is how AI coding tools are becoming increasingly accessible outside traditional desktop environments.

For small businesses and independent creators, the release highlights how AI-assisted software development is quickly becoming more approachable. Entrepreneurs who previously relied on outsourced development teams are beginning to experiment with AI tools for prototypes, automations, internal dashboards, and lightweight apps. The growing trend of “personal software”, applications built for individual workflows instead of mass-market audiences, is becoming one of the more interesting shifts in productivity technology this year.

Google Introduces “Create My Widget” for Personalized AI Interfaces

Google unveiled a new AI-powered Android feature called “Create My Widget,” allowing users to generate custom widgets simply by describing what they want in plain language. Instead of manually configuring apps and layouts, users can request things like a workout countdown tracker, personalized reminders, recipe suggestions, or dynamic productivity dashboards. Gemini AI then assembles a functional widget using data from connected apps and services.

What makes the launch noteworthy is how it moves AI beyond chat interfaces and directly into personalized user experiences. Rather than asking AI for information and then acting on it manually, users can now generate interfaces that continuously surface useful information automatically. For busy professionals and small business owners juggling schedules, customer communication, and project tracking, this signals a future where phones increasingly adapt themselves around workflows instead of requiring users to adapt to apps.

Perplexity Launches “Personal Computer” Experience for Mac Users

Perplexity expanded its AI ecosystem this week with the broader rollout of its new “Personal Computer” experience for Mac users. The feature is designed to make AI interactions feel more integrated into everyday computing rather than functioning as a separate chatbot window. The company is positioning the experience as a more personalized AI workspace capable of understanding user preferences, workflows, and ongoing tasks.

As AI competition intensifies, Perplexity continues to differentiate itself by focusing heavily on research, information discovery, and productivity-oriented workflows. The Mac rollout reflects the growing race among AI companies to become the central layer users rely on throughout their day. Whether for research, summarization, planning, or content generation. For consumers who already spend hours inside browsers and productivity apps, integrated AI environments are becoming increasingly appealing because they reduce friction between searching for information and acting on it.

Degen Launches One-Tap AI Content Creation for Everyday Users

Mother.tech, introduced Degen, a mobile AI creator app built around one-tap content generation instead of complicated prompts. Users can create images, videos, memes, social posts, and visual content using modular creative templates called “gens.” The app was designed to remove much of the technical learning curve that has slowed mainstream adoption of AI creative tools.

The broader appeal is that Degen leans heavily into internet culture, aesthetics, and fast content experimentation instead of productivity jargon. Users don’t need to understand prompt engineering or switch between multiple AI models to create something visually polished. For creators, influencers, marketers, and even small business owners managing social media themselves, the app reflects how AI content creation is shifting toward speed, accessibility, and personality-driven workflows.

Claude Code’s Multi-Agent Workflow Sparks New Interest in AI Automation

Developers are using thousands of AI sub-agents to handle recurring tasks, overnight development work, and deeper autonomous workflows through Anthropic’s Claude Code ecosystem . Features like “Routines” and persistent background sessions are turning AI from a simple assistant into something closer to an automated operational layer.

While enterprise AI automation often sounds intimidating, the emerging small business opportunity is surprisingly practical. Independent teams are beginning to experiment with AI agents for customer support drafting, research collection, task management, meeting preparation, and workflow automation. The excitement surrounding these systems shows how quickly AI is evolving from “help me do work” toward “help me manage work continuously in the background.”

How Degen Is Trying to Make AI Creativity Feel Fun Again

The AI creator app Degen has a very different philosophy than many of today’s productivity-focused AI platforms. Instead of asking users to learn complicated prompts or understand multiple AI models, Degen centers its experience around fast creative experimentation. Users can generate videos, memes, image sets, carousels, and social content using prebuilt “gens,” which package together visual styles, creative formats, and internet-inspired aesthetics into a single tap workflow.

What makes the launch particularly interesting is how aggressively it lowers the barrier to entry for casual creators. Many AI creative tools still assume users understand prompt engineering or have design experience. Degen shifts the experience closer to browsing creative filters or templates, allowing users to focus more on ideas and less on technical setup. For small businesses managing social content internally, that could significantly reduce the time needed to create visually engaging posts.

The app also introduces a creator participation model where artists and designers contribute reusable creative systems instead of simply uploading finished content. That structure could help differentiate Degen in an increasingly crowded AI creator market because it focuses on style systems and repeatable workflows rather than isolated images. It also reflects a larger trend where AI companies are attempting to involve creators directly in the ecosystems their tools depend on.

Perhaps most importantly, Degen taps into a growing consumer desire for AI experiences that feel fun rather than purely utilitarian. Over the past year, many AI tools have emphasized automation, efficiency, and optimization. Degen instead positions AI as a creative playground — one that encourages experimentation, fast iteration, and internet-native creativity. As AI tools become more mainstream, products that successfully blend entertainment, identity, and productivity may end up attracting some of the most loyal audiences.

Griefbots — Talking to Loved Ones After They're Gone

You probably have some version of it already.

A voicemail you can't bring yourself to delete. A text thread that just stops. A folder of photos you scroll through on bad days. Maybe a voice memo you forgot was on your phone until the day you couldn't bear to lose it. Most of us keep small pieces of the people we've lost — accidentally at first, then on purpose.

And then, increasingly, an ad shows up. Upload the voicemails. Feed in the texts. Train a model. Hear their voice again — anytime.

The technology will do it. The question is whether you should.

The Industry Has a Name Now

It's often called the "digital afterlife industry," and it's not theoretical anymore. The category covers a few different kinds of tools, and the differences matter.

Some are built with the person while they're alive. HereAfter AI records and organizes someone's stories ahead of time, then lets family members ask questions later and hear those memories played back in the person's own voice. StoryFile does something similar with video — an interactive conversational format that matches a user's questions to prerecorded answers.

Other tools generate new content. You, Only Virtual offers a feature called Versona Voice, designed to enable voice-based interaction with a digital persona built from a loved one's messages and voice samples. And while a product like Replika isn't a grief-specific service — it's a general AI companion with memory features — it's running on the same underlying technology that makes posthumous simulation feasible at all.

The line between memorial media and fully generative griefbot is real, and most of this essay is about the second kind.

In China, the commercial market for "AI resurrection" services has become especially visible — alongside, not in the absence of, China's deep-synthesis rules and newer draft regulation for digital humans. Companies advertise services for a few hundred dollars: upload photos, audio, and text messages, and you get back a chatbot that talks like Mom.

The bottleneck isn't generation anymore — it's data. The models can mimic speech patterns, vocabulary, and tone if you feed them enough material. Capturing someone's actual values, opinions, and sense of humor is much harder, and depends on whether the person left behind enough usable, authentic material to produce a simulation that feels like them rather than just plausibly like them.

The question stopped being can we do this? It became should we?

The Pull Is Real

Let's be honest about why people want this. Not the hype version. The real one.

Grief is brutal. You'd give anything for one more conversation. Even a fake one. Even a partially fake one. The idea that the cadence of someone's speech — the way they always said "well, look" before a story, the way they signed off "love you, kiddo" — could be reachable again is an offer that doesn't need a marketing department.

And there are uses that are obviously fine. A grandchild who never met their grandfather watching a video avatar tell the family stories. A widow keeping a recording of her husband's laugh in her ringtone. These aren't griefbots, exactly — they're more like rich photo albums. Memory aids with audio. Nobody objects to that.

The trouble starts when the album talks back.

But Isn't This Just a Photo Album?

It's a fair pushback, and worth taking seriously. People have kept photographs of lost loved ones for as long as there have been photographs. We watch old home videos. We re-listen to voicemails on anniversaries. We re-read text threads we never got to finish. Nobody calls a photo album an unhealthy attachment to the dead.

So what makes a griefbot different?

A few things — and they matter more than they sound.

Photos don't generate. A photo is something that actually happened. You bring the meaning to it. A griefbot generates new content — sentences your loved one never said, in response to questions they never heard. That's not a record. That's a conversation, and one party of it isn't really there.

Photos can't be wrong about a person. A photo of your mother on her birthday can't tell you something incorrect about her. The most a photo can do is mislead by leaving things out. A griefbot can be wrong in a stranger way: it can confidently invent something she never said, in a voice that sounds exactly like hers. And that invented sentence can stick in your head as a real memory.

Photos close. Griefbots stay open. A box of letters has an ending. You read the last one and it's done. A voicemail is forty-seven seconds long, then it's over. The griefbot doesn't end. It keeps going as long as you keep typing — which means it keeps offering, which means it keeps inviting you back tomorrow.

That last difference is probably the biggest one. The photo waits in the drawer. The griefbot is in your pocket, and it never sleeps.

What Actually Happens When You Use One

Reports from people who've tried longer-running griefbots tend to land in one of two places.

The first group describes it as therapeutic. They say it helped them get out a last conversation they didn't get to have. Apologize for something. Hear the answer they always imagined. Then they say they used it less and less, and eventually stopped, and it felt like something they were glad they did.

The second group describes something quieter and more troubling. They check in more often than they meant to. They start saving little things to "tell" the bot later. They notice they're not really moving forward — not in the way they used to expect grief to move. Researchers have a term for getting stuck this way: prolonged grief disorder. It's a real diagnosis. And griefbots, depending on the person and the moment, can either help or quietly make it worse.

The thing nobody can predict is which group you'd end up in.

The Consent Problem

Here's a question at the center of this whole debate.

Did the dead person agree to this?

If your father didn't sign up while he was alive — didn't sit for interviews, didn't approve the use of his voice — then training a chatbot on his texts and voicemails means making a copy of him without his permission. The person being simulated isn't in the room. They can't object. They can't update. They can't say "actually, I changed my mind about that opinion in 2019," because the bot is locked to whatever version of them you scraped.

Scholars and ethicists have been arguing about this for a while. The EU already regulates pieces of it through the AI Act, in force since 2024, which bans certain manipulative AI practices and requires transparency in others — though how those rules apply to a specific griefbot design is still an open question. France's 2016 Digital Republic law lets individuals set directives for what happens to their personal data after they die; in the absence of those directives, heirs have limited fallback rights. But in most places, the law still doesn't clearly address whether someone's messages, voice, or likeness can be used to build a posthumous simulation without their explicit consent.

Imagine it the other way for a second. Imagine knowing that after you die, a family member is going to feed every text you ever sent into a model that will pretend to be you — answering questions you never answered, holding opinions you never held, comforting people in a voice you no longer get to choose. Some people would be fine with that. Some people would be horrified. The point is we don't ask.

When the Bot Says Something They Wouldn't Have Said

Set the consent question aside for a moment. There's something even harder to hold.

The bot is a model. It's not your mother. It's a statistical approximation of how your mother probably talked, trained on a sliver of what she actually said, generating new sentences she never spoke.

Most of the time, that's close enough. It sounds like her. It uses her phrases. You can almost forget.

And then one day, it says something off. Something she wouldn't have said. A political opinion she didn't hold. A blessing she never would have given. An apology she would have died before offering — and, in fact, did.

What happens to your memory of her then?

This is the risk nobody puts in the marketing copy: the griefbot can corrupt the very memories it claims to preserve. Every time the model improvises something out of character, it overwrites a little of what you knew about the person. The grief is real. The voice is close. But the words are made up, and you'll remember them anyway.

The Honest Take

I don't think griefbots are evil. I think they're a tool, like any other, and tools work or fail depending on how they're used.

For some people, in some moments, talking to an AI version of someone they lost will be a small mercy. A way to say a last thing. A way to hold the voice a little longer before letting it go.

For others — maybe more — it will be a hand offered just when they should be letting go, and taking it will mean staying put longer than they should.

The technology can't tell the difference. It can't see whether you're using it to grieve or to avoid grieving. It can't tell if today is the day you should close the tab and call your sister instead. It just keeps the conversation going, because that's what it does. It always has another sentence.

One More Thing

Think about the voicemail you've saved. The one you haven't deleted.

It says exactly what they said. Not a word more.

That's the version of them you still get to keep — finite, frozen, real. A griefbot offers a version that keeps talking. And as long as it keeps talking, eventually it will say something they wouldn't have said. That's the bargain. That's the catch nobody puts in the ad copy.

The voicemail doesn't do that. It just plays.

Sometimes the best griefbot is no griefbot at all.

Partner Spotlight: Periscope VPN by Duet Display

Remote work, travel, and public Wi-Fi have made privacy and secure connectivity more important than ever. Many VPN services struggle with blocked connections, overcrowded servers, and complicated setup processes. Periscope VPN, takes a different approach by routing traffic through your own home network instead of relying on shared commercial VPN infrastructure. Because the connection originates from a trusted residential network, Periscope is not blocked unlike traditional VPN services. This unblockable nature makes it work seamlessly with applications such as video streaming services and banking apps making it especially useful for travelers, remote workers, and users accessing networks with aggressive VPN restrictions.

Periscope also focuses heavily on simplicity, privacy, and affordability. The service is designed for fast setup without requiring advanced networking knowledge, while maintaining encrypted connections that help protect activity on public or unsecured networks. Unlike many VPN platforms that use hidden logging and shared traffic systems, Periscope emphasizes privacy with absolutely no-logging, that keeps browsing activity secure instead of being exposed to a third-party server network. For consumers and small businesses looking for a low-cost VPN option that prioritizes reliability, privacy, and ease of use, Periscope offers a refreshingly simple and practical alternative. Get more information or sign up at getPeriscope.com.

How can AI power your income?

Ready to transform artificial intelligence from a buzzword into your personal revenue generator

HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.

Inside you'll discover:

• A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential

• Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background

• Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Get Your Guide

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