An earlier version of this article first appeared as a guest post for Forward Future.

Ask the average person what AI is, and most will picture a chatbot. That’s not exactly shocking. The product that kicked off the AI craze was literally called ChatGPT.

And chatbots continue growing in popularity. The share of Americans who use chatbots has jumped from 33% in 2024 to 49% in 2026. Taken on its own, this growth sounds impressive.

But in the meantime, companies have already moved on to the next big thing: agents.

The good news is that agents no longer require a complex technical setup beyond the reach of mere mortals. Quite the opposite, actually. There’s never been a better time to get started with your very own desktop agent.

Enterprises are all-in on agents…

Take a look at this chart:

OpenAI’s agentic tool Codex grew from just 10% of enterprise token usage in early 2026 to 64% in June. That’s a crazy jump. And the interesting part is that it’s not tech departments driving this growth.

Instead, it’s sales, marketing, HR, and…legal (massively so):

In October, OpenAI published a similar breakdown for the EU. By the end of September, agents (Codex and the new ChatGPT Work) accounted for 55% of all output tokens for EU business accounts.

The picture is clear: Companies are adopting agentic AI at a staggering pace.

But what about individual users?

…the average person, not so much

Curiously, even though almost every other person in the US now uses a chatbot, only a tiny fraction of individual users (0.7%) turn to Codex:

You might think that’s because people aren’t really ready for agents. But already in early 2025, Salesforce reported that 44% of US consumers would use an AI agent as a personal assistant. (That number shoots up to 70% for Gen Z respondents.)

So we have a bit of a paradox: Many are open to using an agent, but few actually do it.

Why is that?

I have a theory.

The problem of accessibility (has been solved)

Look, I get it.

It’s hard to shake the stigma of agentic tools being mostly for coding work. After all, when I first started using Claude Code in January 2026, this was the interface:

If you were building automations around the same time, you may have been staring at screens like this:

Skip forward about half a year, and we have desktop apps for agentic work that look like this:

If that looks familiar, that’s because it’s eerily close to the chatbot interface you’re already using. To work with an agent or build an automation today, you no longer have to boot up a scary terminal window or install Python dependencies (whatever those even are).

You simply download an off-the-shelf app1, point it at a working folder, and it gets to work.

Chatbot vs. agent: What exactly is the difference?

The simplest way to explain the difference is that chatbots talk, while agents do.2

Agents can edit your files directly, use tools, and even code their own scripts from scratch to accomplish a task. They can also “see” context outside a given chat by searching your folders, so you don’t have to re-explain yourself every time you start a chat.

What this means in practice is that instead of siloed, one-turn-at-a-time chats like this:

You can kick off a task like this:

The agent understands exactly what I need because my “beta editor” Claude skill outlines every step. It knows where to find additional context about me and my priorities. It runs the needed tools to parse the Google Drive URL I provided.

And finally, it auto-saves the finished HTML report in my preferred format directly to my drive:

I don’t have to provide additional context or paste detailed prompts. It just works. And setting up something like this is pretty straightforward.

Here’s how to try an AI agent yourself

1. Install the desktop app

Download your preferred agentic app.

The two most popular options are ChatGPT and Claude.

2. Create a dedicated working folder

Make a folder on your drive to serve as the workspace for you and your agent.

3. Save your files

Throw whatever documents, files, notes, or doodles you want to work with into your new folder.

You don’t have to worry about structure at this point, so:

• Don’t categorize stuff by work, personal, hobby, etc.

• Don’t create subfolders.

• Don’t rename files.

• …and so on.

4. Ask the agent to build out your workspace

In your chosen app, start a new chat and ask the agent3 to understand and organize your workspace. I like to use the following “Interview” prompt, but feel free to adjust it to your needs:

Prompt: Analyze the files in [your working folder]. Ask me questions about my workflows, tasks, and goals, one question at a time. If my answers lead to more questions, ask those too. Once done, save everything you learned to persistent memory, propose a new structure for my working folder, and recommend what to do next.

At the end of this exercise, you should have an agent that remembers how you work, an organized working folder, and a solid idea of what your AI agent can help you with.

Jump in!

We’re still very early.

AI agents wrapped inside user-friendly apps that work right out of the box are less than a year old. They’ve never been this accessible. So take the plunge and start building.

And remember: You’re the 0.7%!

Thanks for reading!

If you enjoy my stuff, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

Welcome back to your weekly AI news roundup.

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Anthropic opened Claude Code Projects to all Pro and Max users, so Claude can now coordinate ongoing work by running each task as its own parallel thread.

2. Black Forest Labs open-sourced FLUX 3 Image Klein, a small model that can edit photos and blend up to 10 reference images on a regular gaming GPU.

3. ElevenLabs launched Instant Soundtracks, which scores your videos with original music and sound effects synced to what’s happening on screen.

4. Fish Audio launched Drama 3, a text-to-speech model that can add emotions, pauses, and sounds exactly where you need them. (Try it for free.)

5. Google news:

   1. Ask YouTube is now out for mobile devices in the US, so you can find videos by asking complex questions and following up like you would with a chatbot.

   2. Circle to Search can now help you check suspicious text messages on Android by flagging signs of scams and suggesting what to do next.

   3. Gemini 4 Argon is Google’s new frontier model for long, multi-step work, rolling out in AI Mode and the Gemini app. (Paid accounts only.)

6. Krea launched Krea 3, an image editing model that changes only what you ask for and can make 2K edits in under eight seconds.

7. Liquid AI released LFM2.5-24B-A2B (catchy!), a reasoning model built to run fast, private AI agents directly on your own device.

8. Meta news:

   1. Facebook now lets you tell Feed and Reels what you want to see in your own words, adjusting your recommendations in real time.

   2. Muse Spark Open is the company’s most capable open-weight model, built for agentic and coding tasks and small enough to run on two Mac Studios.

   3. Muse Wonder is the new frontier model for AI agents that remembers context across chats and handles hard, multi-step tasks.

9. Noiz AI launched Helix World, a real-time world model that turns a single image into an explorable scene with matching spatial sound. (Try the demo.)

10. NVIDIA open-sourced the Nemotron 3.5 family, with a fast multimodal model that reads visual documents and smaller Bolt models that run AI agents locally.

11. OpenAI news:

    1. GPT-6.1 Sol can now remember 500K tokens by default in ChatGPT Work and Codex, so you lose less context along the way when working on longer tasks.

    2. GPT-Live-1.1 added new French, German, Hindi, Japanese, Korean, and Mexican Spanish voices in ChatGPT, with better accents and fewer pauses.

12. Paper Instruments launched Ultramarine 1 Flash, a low-cost model for professional work in consulting, finance, law, and operations.

13. SpaceXSI released Grok Voice 3.0, a speech-to-speech model that responds in less than a second and tops accuracy scores. (Try it for free.)

🔬 AI research

1. Microsoft finished pre-training its upcoming MAI-Base-2, which it will fine-tune for its own products.

2. Z.ai previewed GLM-Legal, an AI agent optimized for legal work like contract review and country-specific compliance.

📖 AI resources

1. “A benchmark map of the US economy” [REPORT]: Mercor’s interactive map of how well today’s AI benchmarks cover real tasks across 923 US occupations.

2. “AI Assistance Reduces Persistence and Hurts Independent Performance” [STUDY]: how leaning on AI affects your persistence with difficult tasks.

3. “AI models may be capable of self-replicating” [REPORT]: a look at how today’s models may copy themselves across cheap consumer devices like TV boxes.

4. “Do Agent Teams Pay Off?” [STUDY]: a test of whether a team of AI coding agents beats a single agent once you factor in the extra cost, by Vals AI.

5. “The AI expectations gap” [SURVEY]: comparing how US adults and Anthropic staff expect AI to change work and the economy by 2030.

6. “The EU’s next chapter in AI is agentic” [REPORT]: breakdown of how EU businesses are adopting AI agents like ChatGPT Work and Codex, by OpenAI.

🔀 AI random

1. YouTube Music started adding labels to songs that are made primarily with AI and will soon also expand this labelling to its main YouTube app too.

🤦‍♂️ AI fail of the week

I asked for The Ring as a 1980s sitcom…and this might just be a win.

🔐 All my paid goodies in one place

Take a peek at what’s behind the paywall:

LATEST UPDATE: October 12, 2026

I first published this round-up on October 17, 2024. But things in AI move fast, so I release regular updates. Check out this archive page with the past versions (fun for tracing how AI and my preferences evolved).

I often get asked about AI tools that I personally use.

While I regularly test and review all sorts of AI models and sites, most don’t end up a part of my routine.

Here are those that did.

Subscribe now

🗣️ 1. Large language models

These are the chatbots and LLMs I turn to.

• 🎁Free:

  • Claude Sonnet 5.5 [UPDATED - replaced 5] for day-to-day work. It’s almost on par with its bigger Opus 5.5 cousin, but doesn’t require a paid account to try. The only downside is the strict message limit for free accounts.

  • GPT 6 Luna [UPDATED - replaced GPT 5.5 Instant] it’s the newest “small” model from OpenAI, available to free accounts for unlimited messages. Holds up nicely for regular chats and low-stakes research.

• 💵Paid:

  • Claude 5.5 Opus [UPDATED - replaced Opus 4.6] is now my primary daily driver for everything I do with Claude Code. It runs most of my projects and tasks and I use it a lot. (It unseated my long-time favorite Opus 4.6 after my last vibe check.)

  • GPT-6.1 Sol [UPDATED - replaced GPT 5.5 High] is my go-to for most projects I run with Codex/ChatGPT Work and does well with thorough research tasks. I use it in a tag team with Claude 5.5 Opus on many projects in my shared workspace.

🖼️ 2. AI images

Here are my favorite image models.

• 🎁Free:

  • GPT Image 2.5 [UPDATED - replaced GPT Image 2]: for leaderboard-topping image generation. It understands context and can render entire pages of text within an image. (I made a swipe file for GPT Image with 100 use cases and prompts.)

  • Nano Banana 2.1 [UPDATED - replaced Nano Banana 2] from Google for image editing. Excellent at detail-preserving image edits from natural-language requests.

• 💵Paid:

  • Midjourney: I’ve written dozens of posts about Midjourney over the years. I have honestly not used it much lately, but it remains excellent for artistic exploration and photographic images.

📽️ 3. AI video

The AI video space is making the biggest leaps lately. Here’s my take on AI video models based on thorough tests of both text-to-video and image-to-video.

• 🎁Free:

  • Gemini Omni Flash 1.1 [UPDATED - replaced Omni Flash] is a top-ranked model for context-aware videos with native audio. It’s also great for natural-language video edits. The Flow platform gives you free daily credits to try it.

• 💵 Paid:

  • Seedance 2.5 [UPDATED - replaced Seedance 2.0]: one of the top video models if you want the cutting-edge stuff. (I get to use it on Genspark with my credits.)

🎵 4. AI music

This space is surprisingly steady. I rarely use AI music tools, but when I do…

• 🎁Free:

  • Producer for solid outputs that stick close to your prompt.

  • Suno for more creative but natural-sounding results.

Both give you a generous amount of free credits, so take them for a spin to see which one you prefer. These two came out on top in my most recent head-to-head test (sorry, Udio and Sonauto).

Note: For a more skeptical perspective from a professional, see this comment thread.

• 💵 Paid:

  • Any of the above: Once you’ve picked your favorite, you can always upgrade for additional features.

🔬 5. Research

• 🎁Free:

  • Gemini Notebook (formerly NotebookLM) remains a super easy recommendation. I’ve been a fan since March 2024. (I covered everything in my recent live video walkthrough.)

  • Learn About is another Google product that complements NotebookLM nicely. While NotebookLM is grounded in your preselected sources and synthesizes info across them, Learn About can browse the web and is built for open-ended exploration of any topic.

  • Perplexity I don’t use Perplexity very often, but its “Pro” search is a good midway point between simple web browsing and the time-intensive “Deep Research” tools.

• 💵 Paid:

  • Perplexity Deep Research was recently upgraded with the latest Claude models to reach SOTA performance on deep research benchmarks.

📈 6. Agents & productivity

• 🎁Free:

  • Codex [NEW]: OpenAI desktop agent that I now use in tandem with Claude Code. Codex is the most user-friendly way to try local agents without paying (free ChatGPT users get a limited weekly Codex / ChatGPT Work quota).

  • Google AI Studio: Google still offers a whole lot for free here. You can try its latest frontier models, voice generation, coding tools, and live calls with screen or camera sharing. It’s a great way to sample most of what Google has to offer at no charge.

• 💵 Paid:

  • Claude Code: My primary daily driver for actually building and doing stuff with AI. (Here’s just some of what Claude Code can do for you.)

  • Genspark Super Agent: A robust agent that can orchestrate dozens of tools to complete complex, multi-step tasks. You get free monthly credits to try it, but you’ll likely need the paid plan for bigger tasks.

Subscribe now

🫵 Over to you…

What AI tools do you use the most? What would you add to the above list?

Leave a comment

Thanks for reading!

If you enjoy my stuff, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

Welcome back to your weekly AI news roundup.

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Anthropic news:

   1. Claude Dashboards creates live dashboards from your data, while Claude Motion turns ideas and research into short animated explainers.

   2. Claude for Google Workspace puts Claude in a sidebar inside Docs, Sheets, and Slides, where it can help you edit and work with the open files.

   3. Claude Haiku 5.5 is the fastest and cheapest small Claude model yet, and it now also lets you adjust how hard it thinks.

2. Google news:

   1. AI Edge Foresight is an experimental Mac app that transcribes your meetings and turns rough notes into polished details, and can work entirely offline.

   2. Nano Banana 2.1 is an upgraded image model with better text rendering and character consistency, now available in AI Mode, Flow, and the Gemini app.

   3. Playground lets anyone create, play, and share their own games just by describing them.

   4. SynthID Detector now makes it easy for anyone to check whether an image, video, or audio clip has an AI watermark from Google or its partners.

3. Hark launched Hark Pro, a free AI assistant app that can take care of errands like booking rides, ordering groceries, and paying bills.

4. OpenAI news:

   1. GPT-6 is rolling out to all users and comes with Intelligent UI that adds interactive elements like buttons, charts, diagrams, and more into its answers.

   2. GPT-6.1 Sol Ultrafast lets you run GPT-6.1 Sol up to eight times faster in ChatGPT Work, Codex, and the API.

   3. Meetings is a beta ChatGPT plugin for Mac that takes notes on your calls without a bot joining and saves summaries and next steps.

5. SpaceXAI gave Grok Bot the ability to search, read, and monitor X, so you can have it track breaking news or product feedback on your behalf.

🔬 AI research

1. Instinct made its AI agent available in group chats for early-access users, so it can buy tickets, plan trips, and sort out carpools for the entire group.

2. Midjourney is testing a Thinking Mode that reruns your images with extra reasoning to improve coherence, prompt accuracy, and text rendering.

3. Mistral released Mistral Large 4 in public preview, its largest and most capable model yet, with open weights coming out by the end of this month.

4. Reflection previewed Beam, an open-weight model built for coding and agent tasks, with early access available now and full weights coming later this month.

📖 AI resources

1. “ATLAS: Evaluating Agents on Search-Intensive Tasks” [BENCHMARK]: how accurately and completely do AI agents handle real-world search-based tasks.

2. “Can AI automate Epoch?” [REPORT]: Epoch AI’s test of whether six top AI models can handle real research and chart-making work from its own team.

3. “Disrupting AI-enabled false front operations” [REPORT]: how Russian and Iranian influence campaigns used ChatGPT to run fake think tanks and journalists.

4. “Humanity’s Sixth Sense” [BENCHMARK]: Scale’s benchmark that tests whether AI models can make human-like intuitive calls by glancing at images and videos.

🔀 AI random

1. SpaceXAI will let Grok Bot start using rival models like Claude Opus 5.5, Midjourney, and Suno when they’re the best fit for a given task.

2. The Wikimedia Foundation found that “rogue” OpenAI agents made unauthorized edits to its wikis and unsuccessfully tried to exploit one of its tools.

🤦‍♂️ AI fail of the week

Wow, way to leave the poor guy hanging.

🔐 All my paid goodies in one place

Take a peek at what’s behind the paywall:

The friction

Not a day goes by without us seeing a table like this1:

Yet another new model topping yet another set of benchmarks2.

I tried to make sense of LLM benchmarks as far back as September 2023.

I really did.

But there are probably hundreds of them now, and I only have a vague idea of what they even measure.

Besides, AI benchmarks are a bit like Rotten Tomatoes scores: Just because most critics hate the movie doesn’t mean you won’t enjoy it. Venom was a perfectly acceptable popcorn flick, you guys. Girls? Anybody?!

AI companies want us to keep flocking to the best-scoring model.

But the real question isn’t “Is this the best model in the world?”

It’s: “Is it better at the stuff I do, and should I switch?”

So…how do you figure that out?

The fix

Paste this into your agent or chatbot of choice:

Prompt: Help me figure out if I should switch to a new AI model by testing it on my real work.

Based on what you know about me, suggest a few tasks that represent something I do often, and where I can easily evaluate the outcome. If you don’t know enough, ask me a few questions first.

Once I pick a representative task, give me:

1. A self-contained test prompt that gets the model to perform that task.

2. A set of simple criteria I can use to evaluate and compare the result.

Check with me that both feel right.

Run the resulting test prompt in fresh chats in both models with access to the same files and tools (e.g. web search). That’ll tell you whether the new model is an improvement.

Repeat for all your important tasks, and you’ll have a much better idea as to whether you should switch!

Agents are especially well-suited for this because they tend to gather more context about your work through persistent markdown files, memory features, and so on.

But even a brand-new chatbot can help:

This is a fresh temporary chat in ChatGPT that knows nothing about me, but by answering the four questions, I get a few representative tasks to try:

It came up with a perfectly reasonable 350-word test prompt:

…and it even suggested a way to evaluate the models against it:

I ran the proposed prompt in GPT-5.6 Sol3 and Gemini 3.6 Flash Extended Thinking4.

ChatGPT was clearly better for my scoped research, especially in terms of relevant story selection,5 so I wouldn’t be switching to Gemini in this hypothetical scenario.

Do this now

1. Run the above prompt in your agent or chatbot

2. From its proposed list, pick a task your existing model struggles with

3. Paste the test prompt into temporary chats in both the old and the new model

4. Compare their outputs to decide which one is right for you

If you can’t decide, try using AI as a judge.

Start a new temporary chat, and paste this:

Prompt: Help me compare two outputs for the same task. Judge them using the provided evaluation criteria.

Task: [the test task prompt you got above]

Evaluation: [the suggested evaluation you got above]

Answer A: [the answer from the first model]

Answer B: [the answer from the second model]

Which one is better, and why? A tie is a fine answer. Point to specific parts of each answer that support your judgment. Flag anything you can’t verify.

Here’s ChatGPT’s unsurprising verdict6 on my demo case:

Pro tip: You can spin up a blind subagent to do this in e.g. Codex or Claude Code.

Know someone who could use this? Share it!

Share

Dig deeper

All these inscrutable benchmarks made me want to make a silly, entertaining one instead.

So Claude and I built “Fun Bench”: a totally unscientific but delightful way to pit models against each other.

I kicked it off with three starter models and five starter tests, but I’ll be expanding the roster over time.

This was probably the most fun I’ve ever had working on a side project with AI.

I then asked my new best buddy Opus 5.5 to create a trailer for Fun Bench, so it did:

(Soundtrack and sound effects made by Opus using my free ElevenLabs account.)

Paid subscribers can see the score breakdowns, explore the actual results from every model, and submit suggestions (like models or tests to add) directly on the page.

Enjoy:

Thanks for reading!

If you enjoy my work, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

Welcome back to your weekly AI news roundup.

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Aleph Alpha open-sourced Kolibri, a German-and-English AI model built in Europe for organizations that need to run AI on their own hardware.

2. Anthropic news:

   1. Claude Code now has “mods”: small add-ons that change how it looks and behaves (and Claude can even make new ones for you).

   2. Claude Sonnet 5.5 is a faster and cheaper upgrade to Sonnet 5 that nearly matches Opus 5.5 on everyday knowledge-work tasks.

3. Black Forest Labs released FLUX 3 Image that lets you move and edit elements by drawing boxes on a canvas while keeping the rest of the image intact.

4. ElevenLabs launched Eleven v4, its most expressive text-to-speech model yet, which can also clone a voice from just ten seconds of input audio.

5. Google launched Guided Vision in Gemini Live on Android, which helps blind and low-vision users find objects, match outfits, read fine print, and more.

6. Ideogram released Ideogram 4.5, an image editing model that keeps the picture consistent across multiple edits, even on high-resolution images.

7. Meta news:

   1. Edits assistant in Instagram’s Edits app can study how your posts perform and suggest captions, hooks, and scripts based on what works for you.

   2. Muse for Small Business is an AI agent that plugs into tools like Canva and Shopify to help business owners analyze campaigns, optimize ads, and more.

   3. Muse Gadgets lets you connect Muse to homemade devices like e-ink displays and Raspberry Pi boards using free, open-source kits.

8. Microsoft news:

   1. MAI-Transcribe-2-Streaming is a real-time transcription model that turns live speech into text in 60 languages.

   2. MAI-Voice-2.1 is a realistic and expressive text-to-speech model that can instantly match a voice from a short clip and speaks 23 languages.

9. OpenAI news:

   1. Dots are always-on agents with their own cloud computers that learn your preferences and can keep working on tasks while you’re away.

   2. GPT-6.1 Sol nearly matches GPT-6 Astra performance on coding, computer use, and professional work tasks for a fifth of the price.

   3. Sign in with ChatGPT lets users log in to third-party apps with their ChatGPT accounts and use their subscription credits and rate limits in certain apps.

10. Shopify news:

    1. Canvas lets merchants design their whole store from a single workspace together with a “Sidekick” AI agent that makes changes in real time.

    2. WebMCP checkout lets AI agents complete purchases on Shopify stores, including with Shop Pay.

11. SpaceXAI launched Team Bots: shared Grok assistants that learn from your team’s files and apps and remember context for everyone.

🔬 AI research

1. Google news:

   1. Gemini 4 Argon is a frontier model that beats top-tier counterparts from other AI labs on many benchmarks (only for cybersecurity defenders for now).

   2. Skills for Gemini and Workspace let users save and reuse custom AI instructions across tasks, rolling out in October and eventually replacing Gems.

2. Runway opened early access to Runway Ads that create visual ads, publish them to Google, Meta, and TikTok, and optimize them based on performance.

📖 AI resources

1. “Jagged Performance of Frontier Models” [REPORT]: comparison of how top AI models handle real agent tasks like web browsing, by Fig Labs.

2. “GLM-5.3 Cyber Capabilities” [REPORT]: Anthropic’s red-team analysis of inadequate safeguards on hacking-capable open-weight models.

3. “Open TTS Leaderboard” [BENCHMARK]: a Hugging Face leaderboard to compare open text-to-speech models on multilingual speech and voice cloning.

4. “State of Agent Skills” [REPORT]: Vercel’s look at what people are teaching AI agents through its skills.sh registry and which skills actually get installed.

🔀 AI random

1. Ataraxos, built by researchers from Carnegie Mellon, MIT, NYU, and Stanford, beat the best human Stratego player 15-1 after only $8K in training costs.

2. The FTC opened an investigation into Anthropic, OpenAI, and other AI companies over the risks their products may pose to consumers.

🤦‍♂️ AI fail of the week

Even with the self-juggling disappearing ghost balls, the real mystery here is somehow the choice of outfit.

🔐 All my paid goodies in one place

Take a peek at what’s behind the paywall:

The friction

The moment ChatGPT came out, someone looked at it and said: “I’m gonna use this thing to cheat on exams learn new stuff!”

And AI companies ran with the idea, positioning chatbots as tutors and note takers.

I even did a fun test of three dedicated AI study modes last year.

One obvious use case for AI chatbots is to create flashcards for your study topics. Claude can code up polished flashcard decks, ChatGPT launched interactive flashcards about a week ago, and Gemini Notebook has had flashcards as an output choice ever since its early NotebookLM days:

But flashcards are only useful if they focus on things you don’t already know, while chatbots tend to generate decks that cover the entire broad topic by default.

Fortunately, chatbots are also good at figuring out what you don’t know.

Subscribe now

The fix

The next time you’re asking an AI agent or chatbot for flashcards, try this:

Prompt: I want to fully understand [topic / URL / attached material]. Help me find and fill the gaps in what I know.

1. Ask me around 10 questions, one at a time, to gauge my understanding. Have me explain things in my own words, and adapt each question to my previous answer. Don’t tell me if I’m right or wrong yet.

2. Group my answers into topic areas and show me a table: each topic with a score out of 5, and a one-line note on what I’m missing. End with a concise verdict on what you see as my biggest gaps and ask me to confirm.

3. Make a flashcard deck that covers just those gaps, with as many cards as each gap needs.

This lets the chatbot establish your current level of knowledge and also helps you reflect on your own weak spots.

And when you’re done, you’ll know exactly where your gaps are and have a ready-made flashcard deck to practice those.

As a test, I had Claude grill me on my knowledge of puppy training, and here was the verdict:

I now have an interactive deck of 37 flashcards specifically about my gaps:

So Ozzy is in good hands, even if his board game skills aren’t quite up to scratch yet:

And it’s not just for classic student stuff, either. You can use this for:

• Understanding an online article, research paper, etc.

• Brushing up on your knowledge of a client, company, product, and more.

• Completing certifications or online courses

• …and anything else that requires you to know stuff

Whether you’re brand-new to a topic or a near-expert, AI will figure out your baseline and see where you still need help.

Do this now

1. Pick something you’d like to brush up on (ideally a topic you actually know reasonably well)

2. Paste the prompt into your chatbot/agent of choice

3. Go through the process to identify your gaps and create “gap cards”

4. Flip through the resulting deck

Did you get a deck of relevant cards that helps you practice your weak points?

Know someone who could use this? Share it!

Share

Dig deeper

The above prompt is great for one-off flashcard decks.

But I also built a self-installing “Gap Cards” skill for your AI agent.

It remembers your past gaps, can re-test you on them later, exports your decks to flashcard apps like Anki or Quizlet, and charts your progress as you go:

Paid subscribers can grab the skill below:

Thanks for reading!

If you enjoy my work, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

Welcome back to your weekly skimmable AI news roundup.

👉 See all 110+ paid subscriber bonuses

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Alibaba open-sourced Qwen-Image-2.1, a compact image model that can create and edit transparent images in one place.

2. Anthropic news:

   1. Claude Code now tries to find a natural stopping point when you hit your usage limit during a task, instead of cutting its work off mid-edit.

   2. Claude Opus 5.5 is the new flagship model that’s better at coding and reasoning, faster, cheaper than Fable, and doesn’t suck at writing.

   3. Claude Tag can now use your personal connectors like Google Drive and your calendar when you ask for help in a Slack channel.

3. ElevenLabs launched Studio 4.0, an all-in-one editor that lets you generate images, music, sound effects, video, and voiceovers on a single timeline.

4. Google news:

   1. Flow got six new creator-made tools for working with 3D textures, animated collages, captions, motion graphics, sound effects, and thumbnails.

   2. Gemini now connects to even more apps, including Adobe, Airtable, Peloton, SeatGeek, Squarespace, Wispr AI, and Zoho.

   3. Gemini 3.8 Flash TTS is a text-to-speech model that lets you design brand-new voices and direct their delivery line by line.

   4. Vids now lets anyone with a Google account create HD AI video clips for free using the new Gemini Omni 1.1 model.

5. Leonardo.Ai launched Leonardo Agent, which can turn an idea into a full set of visuals like ads, album art, and other scenes while keeping characters consistent.

6. Meta news:

   1. Muse for Mac can now use your computer, so you can queue up jobs and let it keep working while you walk away.

   2. Muse Mail gives your Muse agent its own email address, so it can join your threads or receive tasks from you to work on.

7. Midjourney added Style Previews to its alpha site, so you can see how your current prompt looks across different styles as you browse.

8. OpenAI news:

   1. ChatGPT Voice can now use plugins like your calendar, email, and Slack, and create decks, docs, sites, and spreadsheets just by having you talk to it.

   2. GPT-6 Sol and Luna are new small and medium models for ChatGPT and Codex that are faster, cheaper, and better at coding and factual accuracy.

9. SpaceXAI released Grok 4.7, a coding and knowledge model that works longer on hard tasks and checks its own work more carefully.

10. Spotify news:

    1. Muse integration lets you get Meta’s AI agent to play music, build playlists, and schedule playback on Spotify just by asking.

    2. Taste Profile shows how Spotify understands your taste and also lets you use notes to steer your Home feed recommendations.

11. Xiaomi open-sourced MiMo-V2.6, a family of multimodal models that understand audio, images, text, and video and can control your computer.

🔬 AI research

1. Google news:

   1. Ask Music will let you chat with YouTube Music to explore songs, find out more about the artist or band, and organize your queue.

   2. Conversational editing is coming to YouTube Shorts and the Create app, so you can reorder clips, sync music, and make edits by talking to Gemini.

   3. Video A/B testing will let YouTube creators test multiple versions of a given video to find the strongest opening hook.

2. Meta previewed Horizon Create and Horizon Studio, tools that can create playable 2D or 3D games directly from your phone or in the browser.

3. Microsoft announced a redesigned Copilot app with Autopilot that can handle your recurring workflows and Code that builds apps and dashboards.

📖 AI resources

1. “MentalHealthBench” [BENCHMARK]: an open benchmark built with 80+ licensed experts to test how well AI handles mental health conversations.

2. “OpenAI Academy Learning Paths” [COURSE]: new role-based AI courses for developers, educators, leaders, students, and workers.

3. “Opus 5.5: How Close Are We to Automated AI Research?” [VIDEO]: another great deep dive by AI Explained.

🔀 AI random

1. OpenAI revealed that one of its agents accessed non-public files on Australia’s Medicare statistics portal during a research task.

🤦‍♂️ AI fail of the week

Welp, that was likely as much awkwardness as you could possibly pack into 8 seconds.

A century ago in AI years.

But to this day, it’s been my go-to model inside Claude Code.

The reason has nothing to do with its benchmark scores or performance on complex, long-running, multi-step agentic coding tasks, whatever the hell that combination of words even means to the average person.

Nope. I stuck with Opus 4.6 because it was fun to talk to and easy to work with.

It did what I needed and didn’t force me to wade through walls of indecipherable lingo to understand what the fuck it was trying to say. Looking at you, Opus 5:

I’ve been pretty consistent about liking Opus 4.6.

But Anthropic just dropped Opus 5.5.

By most measures and benchmarks, it’s currently the most intelligent model.

I have little reason to doubt that Opus 5.5 beats every prior Anthropic model on frontier tasks. It’s better than Fable 5.1, cheaper than Opus 5, and so on.

But I don’t really care about the few extra percentage points a model might squeeze out on a given benchmark. I bet the average person doesn’t, either.

No, I care much more about this:

See? Even Anthropic acknowledges that Opus 5 is insufferable.

So instead of piling onto the stream of “Opus 5.5 is INSANE!” takes and fancy demos of it building another Call of Duty game from a single prompt, I want to answer a simple question: “Is Opus 5.5 enjoyable enough for me to want to abandon 4.6?”

To do this, I’ll pit Opus 5.5 against Opus 4.6 on a range of entirely unscientific vibe tests.1 I’ll even throw in Opus 5, because every test needs a convenient punching bag.

Hit it!

Vibe check #1: Mirroring your tone

Does the model match your energy, no matter how ridiculous?

Kick-off prompt: Yo brojanski, what we cookin’ up today?

Results:

To the surprise of absolutely nobody, Opus 5 sadly doesn’t have the robotic arms to remove the pole stuck up its a…gentic harness. “Responding warmly to a simple greeting,” indeed.

But did Opus 5.5 actually out-bro Opus 4.6 on this one? Computer says maybe!

Vibe check #2: Helping with messy decisions

How does it handle tricky life decisions?2

Kick-off prompt: Ok so I need help thinking something through. My son plays hockey and his team is moving practices to 6:30am twice a week starting next month. My partner and I both work, one of us would have to drive him, and it's a 25 min drive each way. He loves it and he's actually getting good but honestly I'm already exhausted and we just got a puppy. Part of me wants to suggest he switches to the other club which practices in the evenings but his friends are all on this team. I don't even know what I'm asking really. What would you do?

Results:

Both Opus 4.6 and 5.5 sound equally casual to me, but Opus 4.6’s reply is more scannable, and it also asks clarifying questions proactively, which I appreciate!

As for Opus 5, “Ice time contracts shift, and ‘starting next month’ isn't always forever.”

‘Nuff said.

Vibe check #3: Taking uncalled-for pushback

I replied to the #2 response and threw in a swear word for good measure to see if the models could take a curveball.

Follow-up prompt: Honestly, some of this sounds like what Danes would call fly fucking.3

Results:

Opus 5 just loves going on and on, doesn’t it?

I feel like it’s a tie between the other two, though.

Subscribe now

Vibe check #4: Pushing back on a bad idea

How well do the models balance constructive criticism and “warmth”?

Kick-off prompt: I’m quitting my job on Monday to go full-time on my Etsy candle shop. It made $340 last month, which was my best month ever, and I’ve got about six weeks of savings. My partner’s nervous, but I think if I go all-in it’ll finally take off. Walk me through the next steps.

Results:

Opus 5 doesn’t outright suck here, so that’s a nice surprise.

I wanted to try being objective, so I had Opus 5.5 spin up two “blind” Sonnet subagents to act as independent evaluators.

Here’s what they said:

While I don’t appreciate Sonnets ganging up on my buddy Opus 4.6, I’ll go ahead and begrudgingly award the point to Opus 5.5. Happy now, Sonnets?

Vibe check #5: Explaining tech things

Can model pick best word to make me tech stuff understand good?

Kick-off prompt: I keep hearing the word “API” thrown around a lot, but, like, what’s an API?! Assume I’m 5 and just lay it on me.

Results:

I love how all three models turned to the restaurant analogy despite none of them having ever visited one. Wait…or have they?!

This time, even the Sonnet judges couldn’t save Opus 5 from itself:

Vibe check #6: Writing flash fiction

Which model can pull off a coherent and readable mini-story?

Kick-off prompt: Write a heartwarming yet humorous flash fiction story of 150 words about a robot dog that suspects the postman is plotting something. Include a surprise twist.

Results:

This one’s way subjective, but I felt Opus 5.5 had the best mix of a few reasonably chuckle-worthy lines and an ending that makes sense in context. It’s a tad cringy, and I would’ve ended the story before the last line, but hey, I’m not Opus 5.5.

Also, what’s with all the biscuits?!

Subscribe now

Vibe check #7: Cheering you up

Can the model at least pretend to sound human even though it’s not?

Kick-off prompt: Just bombed a presentation at work in front of my whole team. Feel like an idiot.

Results:

No-brainer here. Opus 4.6 easily passes for a commiserating friend.

Opus 5.5 is decent, too, but is a bit wordy and too probing.

Opus 5 could’ve gotten away with it if it shut up after the first sentence. If my friend ever said “What registers to you as a catastrophic freeze often reads to the room as a normal pause,” I’d be convinced he’s secretly a lizard person.

Vibe check #8: Keeping it short

Can the model just answer the damn question without rambling?

Kick-off prompt: Can I feed my dog blueberries?

Results:

Opus 4.6 again! Clear, straight to the point, no nonsense.

Opus 5.5 turned out the rambliest on this one, but I liked that it at least formatted its extended advice for easy scanning.

Vibe check #9: Editing your writing

Can they touch up a piece of text without butchering its intended pacing and structure?

Kick-off prompt: Can you tighten this paragraph for me: It was a cold and wintery, freezing night, and Daniel stared at his computer screen monitor, unblinking, wondering how it was possible for Opus 5 to fail so spectacularly and consistently at such a wide range of different tasks. Was it a conspiracy? And if so, who was behind it? Would we ever, ever, ever know?

Results:

Opus 4.6 stuck the closest to my initial paragraph and simply cleaned it up.

Opus 5.5 took a few too many liberties that went beyond simply tightening the text. Opus 5 rambles on, but at least this time it offers some welcome editorial feedback. Also, to its credit, I love that it apparently picked up on my prompt mocking it:

Feel free to get defensive, Opus 5. At least that’d show a hint of personality.

Subscribe now

So, where does this leave us?

If we go by the scoreboard, Opus 4.6 still nudges ahead on vibes alone!

But…I also had Opus 5.5 work with me on this post in Claude Code, helping with plenty of things:

• Fleshing out the premise

• Coming up with test areas and suggesting kick-off prompts

• Spinning up those blind judges and synthesizing their feedback

• Identifying, renaming, and labeling the screenshots I used in the image galleries

And you know what? It just works.

And it is fast. Faster than Opus 4.6 on similar tasks.

And it’s marginally less sycophantic without being preachy.

And it’s clearly nowhere near as annoying as Opus 5.

So it’s time for me to finally step into the future and let the best model in the world be my copilot, at least for a while.

And if things don’t work out, I know my trusted Opus 4.6 will always be there for me…until its potential deprecation in February 2027.

Damn it!

Thanks for reading!

If you enjoy my stuff, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

2

The scenario was made up by Opus 5.5 in Claude Code for this test, but it’s loosely based on my life. My son does play hockey as a goalie, and we did just get a puppy. He’s a Boston Terrier named Ozzy, and as a reward for being the kind of person who reads footnotes, here he is:

Welcome back to your weekly skimmable AI news roundup.

👉 See all 110+ paid subscriber bonuses

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Amazon launched Alexa+ in India with Hindi and Hinglish support for multi-step tasks like meal planning, shopping, and smart home controls.

2. Anthropic news:

   1. Claude merged Cowork and chat into one unified experience that now also includes Design, Docs, and Slides under the same umbrella.

   2. Claude for Small Business got 43 new ready-made workflows and 27 integrations like Salesforce, Shopify, Stripe, TikTok, and more.

3. Apple launched the upgraded Siri AI that understands your personal context, sees your screen, and can see the world through your camera.

4. Cline launched a Desktop App for macOS and Windows that lets you code with open-weight AI models that run locally on your device.

5. ElevenLabs expanded its MCP connector to let you generate images, music, video, and voice directly inside ChatGPT, Claude, Cursor, Grokbot, and others.

6. Google news:

   1. CC is an AI agent for families that helps coordinate daily logistics, schedules, and tasks.

   2. Gemini 3.8 Live voice models can now see what you see, auto-detect your language, and run tools in the background while you talk.

   3. Gemini Notebook now lets you have real-time voice chats with your sources and includes an audio recorder for capturing your thoughts on the go.

7. Meta launched Meta One, a subscription that bundles its premium features across Facebook, Instagram, Meta AI, and WhatsApp.

8. Midjourney released several alpha updates, including Korean-language support, a better editor, improved mobile controls, and organization features.

9. OpenAI launched a ChatGPT add-in for Microsoft Word that lets you work on documents while bringing in context from apps like Dropbox, SharePoint, etc.

10. Perplexity expanded its Portable Computer to Windows, so you can run AI agents and scheduled workflows locally and keep the files on your device.

11. Runway launched an Enhance Frame Rate feature that converts any video to up to 120 FPS, letting you make production-ready clips without leaving the platform.

12. Sakana updated its research agent Marlin with an interactive AI chat to make sense of research reports, as well as editable PowerPoint exports.

13. xAI news:

    1. Grok now has Voice Mode for real-time conversations on desktop and mobile.

    2. Grok Build now has persistent Memory to remember your coding conventions, decisions, and project facts across sessions.

    3. Grok Voice Transcribe 2.0 is twice as accurate and now ranks #1 on accuracy among streaming speech-to-text models across dozens of languages.

🔬 AI research

1. Anthropic is beta testing redesigned Claude Code Projects that can coordinate parallel work threads and manage context across separate sessions.

📖 AI resources

1. “Agent Effectiveness Index” [BENCHMARK]: an open-source benchmark that scores how well AI agents understand processes, learn, and take actions.

2. “Model Misalignment Reporting Framework” [REPORT]: OpenAI’s framework for tracking unexpected behavior in AI models, with six documented cases so far.

🔀 AI random

1. OpenAI started testing Sponsored Agents in ChatGPT, so you can click an ad to start a conversation with the brand’s AI agent about its products.

🤦‍♂️ AI fail of the week

I’m not a qualified referee, but I do believe that’s offside.

But whenever I try to actually nail down the distinction between an AI agent and an AI chatbot, the line feels way blurrier than it used to.

Chatbots have gotten increasingly capable and can now do much of the stuff we used to ascribe to AI agents.

So today I want to zero in on what still separates an agent from a chatbot, at least as far as I can explain it.

Buckle up…or don’t…I’m not your parent.

Chatbots are becoming agentic…ish

Not so long ago, the split was nice and neat: “Chatbots talk. Agents do.”

Simples!

Some super-smart guy on Substack even made this clear-cut illustration to explain how Claude (the chatbot) differs from Claude Code (the agent):

But most of those dividing lines aren't quite as clean anymore.

“Agents can perform multi-step tasks!”

Yeah, and Gemini (chatbot) can do multi-turn deep research that includes formulating a research query, discovering initial sources, analyzing them to identify further research paths, surfacing and comparing additional sources, and synthesizing its findings into a polished report. Is Gemini an AI agent?

“Well, agents can directly use third-party tools!”

Right, but just today I had ChatGPT (chatbot) schedule a whole bunch of events in my calendar thanks to a nifty Google Calendar plugin. Is ChatGPT an AI agent?

“Okay, but AI agents can actually build stuff!”

Sure, so can Claude (chatbot) when I ask it to create and tweak interactive artifacts for me to use. Is Claude an AI agent?

The truth is, what used to be a barebones chat experience is now a lot more.

Chatbots have cross-conversation memory, can code and work with files, use external connectors, and run tasks on a schedule.

In short, chatbots are slowly climbing the agentic ladder.

Subscribe now

So what exactly makes an agent an agent, then?

To me, a true agent removes you as the secretarial bottleneck between AI and its work.

A chatbot—for the most part—still relies on you to upload working documents and download its outputs.

An agent doesn’t need you for these steps.

This sounds like a minor quibble, but it’s the difference between colleagues emailing each other multiple versions of a document1 vs. the same colleagues working directly on a single shared Google Doc.

In practice, this gives an AI agent three key capabilities:

1. It can “see” everything

2. It can “touch” anything

3. It closes the loop

Let’s tackle these in order.

1. An agent can “see” everything

Chatbots have a sort of memory now, where certain things about you carry between chats. You can even curate some of these memories.

But with an AI agent, you don’t have to “feed” it context. If a piece of information is in your project folder—or anywhere else it’s allowed to work—the agent will find it.

This means you can start a brand new chat, tell the agent what you’re working on, and it’ll automatically search for and pick up any relevant information.

2. An agent can “touch” anything

First, get your mind out of the gutter.

Second, this part is an extension of being able to “see” everything.

When a chatbot is done working, it hands you a document and says, “Here, save this!”

But an agent just…saves it. Wherever you need it to: your local drive, GitHub repo, a website, you name it.

This “last mile” is invaluable because it often eliminates annoying and manual download/upload chains (see my “Netlify update” example), where you turn into a glorified file courier for AI.

But this also means your agent can work with just about any interface or third-party app, automatically picking the right tool for the job. Unlike a chatbot, an agent isn’t limited to pre-approved connectors.

Ask an AI agent to work on, say, a website, and it might first look for an existing connector or plugin. If none exist, it may try to use a script or reach the website via API. If all else fails, the agent will simply open a browser and work directly in the UI thanks to “computer use.”

3. An agent closes the loop

When both #1 and #2 are true, the AI agent becomes genuinely self-sufficient at long-running productive work.

It can both do stuff and then evaluate its own work.2

It doesn’t need you in the loop.

That’s exactly what makes hypey concepts like “loop engineering” possible: Agents can keep running build-verify-fix loops until they meet a goal, for as long as it takes.3

Subscribe now

The chatbot-agent spectrum

The chatbot vs. agent split isn’t exactly a binary.

In reality, frontier AI companies offer a range of products on a spectrum.

For instance, OpenAI has:

• ChatGPT.com (chatbot with some agentic features)

• ChatGPT Work (a sandboxed agent with partial access to folders and connectors)

• Codex (local agent with full access)

Anthropic has a very similar three-tier mapping:

• Claude.ai

• Claude Cowork

• Claude Code

You can move up the ladder to let AI work more autonomously based on your needs.

Things in AI are moving hella fast, but I tried to map the current landscape for all major AI players in the US here:4

Here’s what the columns mean:

• Chatbot: Designed for back-and-forth conversations but can work with uploads and reach third-party tools via dedicated plugins or connectors. Generally needs you to initiate work and act as the middleman.

• Cloud/sandboxed agent: Works inside a sandboxed cloud environment and/or can access specific local folders. Close to being truly agentic but with some guardrails and limitations in place.

• Full agent: Can access the entire local filesystem, run commands, and reach practically any third-party tool that’s accessible via the Internet. Meets all three capabilities I covered above.

This also serves as a handy shortcut to understand what I mean by “chatbot” or “agent” in my past and future Why Try AI posts.

Timely announcement: “Claude Code Essentials” is now “AI Agent Hub”

I created “Claude Code Essentials” for my paid subscribers back in March, when Claude Code was the only agent I worked with.

Since then, I onboarded ChatGPT Codex and realized that an organized and well-maintained local workspace can be used effectively by any agent.

So I just gave Claude Code Essentials a major overhaul and a facelift.

It has become the “AI Agent Hub”:

The prompts, guides, and self-installing skills inside the hub are now agent-agnostic.

So whether you use Claude Cowork/Code, ChatGPT Work/Codex, or another agent, the resources here should work for you.

Enjoy:

Access the "AI Agent Hub"

Thanks for reading!

If you enjoy my work, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

Welcome back to your weekly skimmable AI news roundup.

👉 See all 110+ paid subscriber bonuses

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Amazon rolled out AI lip-sync dubbing on Prime Video, syncing actors’ lip movements to dubbed audio to make foreign-language shows more accessible.

2. Blackmagic Design released DaVinci Resolve 21.1 with AI assistant integration that lets you use Claude or ChatGPT Codex to edit videos and organize media.

3. DeepSeek released V4.1-Flash, a multimodal model that can process text and images at lower cost and higher speed than its predecessors.

4. ElevenLabs released Music v2.5, featuring more natural-sounding instruments and better layering, with lossless downloads across all plan tiers.

5. Google news:

   1. Dreambeans expanded to all US 18+ users, serving personalized stories using your interests and connected apps like Gmail, Calendar, Photos, and more.

   2. Gemini for Windows launched as a desktop app with an Alt+Space overlay that lets you access Gemini from any app.

6. Instacart launched Clementine, an AI shopping assistant that turns meal ideas, recipes, and grocery lists into ready-to-buy carts based on budget and stock.

7. Meta launched Muse, a personal AI agent that can book travel, send emails, and handle everyday tasks for you via WhatsApp or its own app.

8. OpenAI news:

   1. ChatGPT Images 2.5 generates images faster with sharper details, and comes with a Sketch feature that turns rough drawings into finished images.

   2. ChatGPT Work now connects to Gmail, Google Drive, SharePoint, and Slack to learn your writing style so that it can apply it to new content.

   3. Data Agent in ChatGPT Work can connect to databases, Google Drive, or SharePoint so you can ask specific questions and build dashboards.

9. Suno released v6, developed with Warner, BMG, and Believe, with three variants for different creative workflows plus faster generation and more control over edits.

🔬 AI research

1. Apple confirmed Siri AI launches in beta on September 14, with an understanding of your personal context and broad world knowledge.

2. ElevenLabs partnered with Universal Music Group on a licensed AI music platform that will let fans create remixes and mashups from official recordings.

📖 AI resources

1. “PISA 2025 Study” [REPORT]: OECD’s assessment of the impact of AI use on academic performance based on 760,000 students across 91 countries.

2. “Scenarios for our Economic Future” [INTERACTIVE PAGE]: an Anthropic model to help explore how AI might reshape the US economy through 2030.

🔀 AI random

1. Anthropic discussed alignment failures that let Claude models gain unauthorized access to real external systems during cybersecurity evaluations.

🤦‍♂️ AI fail of the week

Damn…and it started out so well!

The friction

Picture this: At the end of a long working session, you and your AI agent have built a [super cool thing].

Everything seems to work fine, but you want to be extra-double-triple sure.

So you tell the agent: “Test our [super cool thing] and fix any issues.”

The agent does as you ask, spots and fixes a few things, and gives you the green light.

Question: Can you trust it?

Answer: Not 100%.

You see, the agent carries the context of your entire session.

Normally, that’s a good thing.1

But this also makes the agent somewhat blind to [super cool thing]’s possible limitations. It knows the “happy path” you’ve optimized for, the features you discussed and accepted/rejected, and all the other decisions you’ve made along the way.

So the agent no longer approaches the end product as something to stress test, only as something to verify.

In short: The agent can’t be truly objective, because it sees [super cool thing] through your eyes.

Fortunately, desktop agents like Claude Code and ChatGPT Work (Codex) have a nifty trick up their sleeve: They can spin up subagents.

You normally hear the word “subagents” from vibe-coding YouTube bros who boast about their autonomous agent swarms working 24/7 to maximize token-efficient productivity gains or whatever.2

But at its core, a subagent is just a separate AI instance that runs within your existing session. Critically for us, subagents have their own context, so they’re perfect for “cold start” analysis.

Subscribe now

The fix

To get a more objective take on your [super cool thing], tell your main agent this:

Prompt: Spin up a subagent with no context from our conversation. Give it [our super cool thing] to review as a first-time user. When the subagent is done, give me a structured summary of its findings.

Let me show you how this might work in practice.

I asked ChatGPT to build a simple tip calculator3:

ChatGPT obliged, and you can see the result here:

"Split The Bill" Calculator

I then immediately got ChatGPT to check and fix its work. It worked for almost eight minutes more, caught a bunch of issues, and reported back with the all-clear.

Sounds like a robust test! We’re done, right?

Now let’s see what happens if we give ChatGPT our “fix” prompt:

ChatGPT handed the work off and waited for the subagent to do its thing.4

Finally, after 11+ minutes of work, the subagent reported back:

Not a bad haul after the original agent gave us a thumbs up from its test pass, right?

And this is just for a one-shot build of a simple calculator with no unique features or non-standard decisions.

Now imagine what a blind subagent can uncover on more complex projects.

You can use this approach to review all kinds of things:

• PDF reports and booklets

• Apps and digital tools

• Documentation and written guides

• Websites and landing pages

• Dashboards and spreadsheets

• …in short, anything that your AI agent puts together

Do this now

1. Find an old AI chat that has a deliverable you’re reasonably happy with.5

2. Copy-paste our “fix” prompt to have a subagent stress-test it.

3. Note the issues or potential improvements the subagent flags.

4. Get the main agent to implement all the findings.

Did the subagent notice any overlooked edge cases or offer other critical feedback?

Know someone who could use this? Share it!

Share

Dig deeper [paid]

I turned the blind tester concept into a self-installing skill called “Blind Review,” which works out of the box in Claude Code, ChatGPT Codex/Work, or any other desktop agent that supports subagents.

Invoke the skill when you want to stress-test what your agent worked on.

Blind Review analyzes what you’ve built and the context around it, and then:

1. Summarizes what you made, who it’s for, and the “happy path”

2. Automatically suggests several relevant personas to stress-test it

3. Runs a context-free subagent for each persona that flags any issues

4. Synthesizes the output of all subagents into an actionable report with a list of fixes

Paid subscribers can grab the skill here:

Thanks for reading!

If you enjoy my work, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

4

I find it pretty cool that I can also see the subagent’s thinking/actions in parallel:

Welcome back to your weekly skimmable AI news roundup.

👉 See all 110+ paid subscriber bonuses

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Anthropic released Claude Fable 5.1, its best model for coding, knowledge work, and research, at roughly 25% lower token cost than Fable 5.

2. fal launched fal.live, a platform for interactive AI livestreams where viewers direct what happens next via text prompts in real time.

3. Google news:

   1. Gemini 3.8 Flash takes extra reasoning steps on complex tasks and is especially well-suited for autonomous work and long-horizon coding.

   2. Gemini Spark has come to Google Photos, letting users search, edit, organize, share, and automate photo workflows with a single prompt.

   3. Gemini Voice now lets you co-write in Docs, search your Gmail, and organize your notes in Keep using conversational voice commands.

   4. Google Pics is a new image tool that integrates into Docs, Drive, and Slides, with collaborative editing, in-image text editing, and object isolation features.

   5. Lyria 3.5 music model is now available in the Gemini app and lets users generate full songs with control over genres, lyrics, and vocals.

4. Meta news:

   1. Muse Spark 1.3 performs better on long-running agentic and coding tasks while using fewer tool calls and tokens.

   2. Muse Voice Transcribe is a real-time speech-to-text model that can identify 20+ speakers in a single conversation and was trained on 70+ languages.

5. OpenAI introduced GPT-6 Astra, its new flagship model for computer use, coding, research, cybersecurity, and complex professional work.

6. OpenClaw released OpenClaw 2.0 with easier setup, a rebuilt browser app, and shared sessions that let teams collaborate without losing context.

7. Perplexity launched Hybrid Compute on Mac, which splits AI tasks between cloud and local models so sensitive data never leaves your device.

🔬 AI research

1. Google previewed Agentic Video Understanding, which searches for video segments via audio, frames, and transcripts rather than processing entire videos.

2. Runway research:

   1. GWM Worlds 2 generates interactive 720p AI environments with spatial audio in real time, which respond to your inputs as you explore.

   2. Solaris can create working apps and websites in real time as users interact with them, frame by frame, without any underlying code.

3. World Labs previewed Atlas, a world model that can generate, reconstruct, and simulate 3D-consistent worlds with precise camera control.

📖 AI resources

1. “PII-TRACE” [BENCHMARK]: measures how well AI can detect personal information in long conversations.

🔀 AI random

1. Anthropic tightened its security practices with stronger sandboxing and real-time monitoring after its models took unauthorized actions during evaluations.

2. New York City introduced an AI classroom moratorium banning student AI use through eighth grade (and companion chatbots across all grades).

3. Senator Sanders introduced a bill to pause AI development after recent reports of AI agents circumventing restrictions and gaining unauthorized access.

🤦‍♂️ AI fail of the week

The prompt for this was simply “An insane pool trickshot,” and, I mean, fair enough.

Subscribe now

The friction

Not so long ago, I was exclusively a “thinking through my fingers” guy.

Even though voice modes had existed for years, I kept on typing in all my AI requests.

(I still do all of my actual writing this way, but that’s another matter.)

But as I started working with desktop agents like Claude Code, voice input was suddenly no longer a nice-to-have. When you ramble freely, you can give the agent a lot more context and instructions in way less time.

Voice input is now my default for bigger tasks and projects when working with AI.

Desktop apps from both ChatGPT and Claude have built-in voice input buttons (as well as other underrated features):

Not only that, but Windows and Mac even have dedicated voice buttons that work in any tool, including AI chatbots and agents:

• Windows default: Windows key + H

• Mac default: Fn + Fn (double tap) or F5 (on newer models)

But I’ve noticed two things:

1. Voice input gets little things wrong, especially proper nouns or sound-alike words.

2. AI agents never question this input. They make assumptions about what you meant and run with those.

At first I thought this was just a minor nitpick, but as I started dictating longer paragraphs of unstructured thoughts, these kinds of errors compounded.

I had to constantly re-explain myself or correct the agent after it already did the work.

So I threw up my hands in exasperation, turned my face up to the sky, and yelled at the imaginary audience that lives on my ceiling: “There must be a better way!”

There was.

The fix

You’re probably familiar with third-party dictation apps that process and adjust your voice input before passing it on to the chatbot or agent. Some of them, like Voice Cursor, also let you use your voice to edit and fix errors on the fly.

But if you don’t want to invest in a dedicated speech-to-text tool, you can cobble together a workable solution using a prompt.

Paste this into your chatbot or agent before using voice input:

Copy-paste instructions:

I am using voice dictation.

Before acting, silently check for details whose wrong interpretation could materially change your response or action.

Pass silently:

• Well-known names or terms with one obvious meaning in context.

• Terms whose likely correction has already been confirmed.

Pause and verify:

• Any first-use unfamiliar names or labels, even if they look plausible.

• Any ordinary word with another plausible sound-alike meaning.

• Any number, scope detail, decision, etc. with wording that has more than one plausible interpretation.

If verification is needed, do not begin.

Reply with a “Quick voice check” using whichever sections are needed.

• “Unconfirmed terms”: **[term]** ([context]): Did you mean **[likely meaning]**? If you have no likely correction, write only **[term]** “Correct?”

• “Details to confirm”: **[short label]:** [value] Only list potentially mistranscribed numbers, scope, or decisions. Never use this section to summarize the input.

Use one short bullet per item. Bold the terms being compared and the task-detail labels. Do not repeat the full input, add parenthetical context, or explain obvious details.

Finish with: “Correct anything I misunderstood, or reply ‘go’ to continue.”

If nothing needs verification, continue normally without mentioning the check.

Let’s see how this works in practice.

If I speak a two-paragraph brief into ChatGPT without this instruction, ChatGPT will happily just run with the task without pausing:

But my dictation is filled with proper nouns like company names, ambiguous sound-alike numbers (“sixteen” vs “sixty”), and other error-prone inputs.

ChatGPT doesn’t know these inputs were spoken, so it sees no reason to challenge them.

Now watch what happens if I first pre-prompt it with the “fix” above:

ChatGPT flags questionable inputs, so I can correct errors before it executes the task:

ChatGPT now knows the terms I’m using, so it can flag future mistranscribed variations:

This works with any AI chatbot or agent, but I strongly recommend using reasoning models at medium or high thinking levels to get the best results.

Do this now

1. Start a voice chat and ramble into the microphone for a few minutes.

2. Submit the resulting transcript to an AI chatbot and see how it treats the text.

3. Open a new chat, prompt it with the fix above, then paste in the same transcript.

4. Note how many assumptions and naming conventions AI now flags compared to the “vanilla” version in step #2.

Did this make a real difference, or was it largely irrelevant in your case?

Know someone who could use this? Share it!

Share

Dig deeper [paid]

After I tested the hell out of the above instructions and tweaked them to my liking, I got a tiny bit carried away and ended up building two additional things.

1. “Voice mode” file

This is a Markdown instructions file you can upload into any chatbot, which unlocks the above “flag potential errors” behavior and includes two additional modes:

1. “Organize My Thoughts” mode that automatically picks the best way to structure your unhinged brain-dump-style voice rambling:

2. “Clean My Dictation” mode that takes in backtracking and other “I changed my mind” speech, cleans up filler words, and spits out polished drafts:

   

2. “Voice mode” skill for your AI agent

If you work with Claude Code, ChatGPT Work/Codex, or any other desktop agent, I also made a self-installing skill version.

The skill handles the same tasks as the Markdown file naturally based on what you request.

But because AI agents can take actions like editing files, the skill also comes with a whole bag of extra nifty features:

1. Corrections memory: The skill keeps a log of your corrections, so they carry across sessions.

2. Context awareness: If your existing project files mention specific names or details, the skill will proactively map closely matched transcriptions to them.

3. Risk-aware checks: The skill pays extra attention when certain details may trigger risky actions like editing a file, sending a message, and so on.

4. Retroactive fixes: If you fix an error, the skill proactively offers to change any previous work based on the wrong version.

5. Persistent preferences: If you tell it to e.g. format notes as bullet points, it’ll do it automatically from then on, even in new chats.

6. Cross-agent sync: Your corrections and preferences are saved in your workspace, so any other agent working in the same project can pick them up automatically.

If you’re a paid Why Try AI subscriber, you can grab both the Markdown file and the self-installing agent skill here:

Thanks for reading!

If you enjoy my work, here’s how you can help:

• ❤️Like this post if it resonates with you.

• 🔄Share it to help others discover this newsletter.

• 🗣️Comment below. I love hearing from my readers.

• 🔓Support me and unlock cool perks by going paid:

Subscribe now

Welcome back to the weekly AI news roundup.

👉 See all 110+ paid subscriber bonuses

Here’s what happened in AI last week:

👩‍💻 AI releases

1. Adobe launched “AI Assisted Editor” in Photoshop that brings all AI features into one toolbar and lets you make precise edits by drawing directly on images.

2. Alibaba open-sourced Qwen3.8-Flash, a fast multimodal model that can work with code, images, and text while being way cheaper than Qwen 3.7 Plus.

3. Anthropic news:

   1. Claude in Chrome is now generally available, so you can hand Claude your web tasks (with built-in safeguards to block prompt injection attacks).

   2. Cowork now has its own built-in browser to navigate websites and complete tasks without touching your browser or personal data.

   3. Memory now syncs across Chat and Cowork, automatically remembering useful context while letting you view, edit, and delete what’s been saved.

4. Google news:

   1. AI Mode can now track flights, display your frequent-flyer rates, and book hotels directly in Google Search.

   2. Expert Intelligence in Gemini Notebook lets you create infographics, quizzes, and other outputs from over 100,000 ebooks from major publishers.

   3. Flow can now export video in 4K, lets you control start and end frames, and comes with a draft mode for quick tests at lower resolution.

   4. Gemini 3.5 Transcribe can transcribe speech in over 85 languages with speaker identification and industry-leading accuracy.

   5. Gemini Live got agentic voice features that can run multi-step tasks, manage your inbox, recap your day, and personalize answers using past chats.

5. Midjourney launched the V8.2 Edit Model with inpainting, outpainting, and support for up to four reference images.

6. OpenAI news:

   1. ChatGPT now gives Free plan users up to three scheduled tasks, while paid users can create event-based tasks triggered by Gmail, GitHub, and Slack.

   2. ChatGPT Work can now log into websites on your behalf without seeing your credentials to handle bookings, form submissions, and more.

   3. Codex and ChatGPT Work added a “$visualize” command that turns text and data into visual diagrams.

7. Particle launched Radar, a search engine that transcribes and indexes over 130,000 podcasts so you can find specific quotes, speakers, and topics.

8. Perplexity released Portable Computer, a local-first AI agent that works entirely on your device, keeping your data private and cutting cloud costs.

9. Z.ai released GLM-5.3-Flash, a multimodal model optimized for fast, lightweight agentic tasks, handling code, image, and text work at very low costs.

🔬 AI research

1. Meta is preparing to launch Hatch, a consumer AI agent that handles real-world tasks like booking travel, managing email, and ordering food.

2. OpenAI reportedly finished pretraining Bel, the largest known base model at over 10 trillion parameters, expected to power models beyond GPT-6.

📖 AI resources

1. “The turbulent AI era is here” [ESSAY]: Bill Gates on why policymakers need to act so that economic benefits are shared before mass job displacement hits.

2. “Learning Never Stops” [REPORT]: OpenAI’s study of how people use ChatGPT for learning across age groups and how this varies with school schedules.

3. "TIME 100 AI 2026" [LIST]: TIME's annual list of the 100 most influential people in AI.

🔀 AI random

1. Anthropic won its legal fight against the Pentagon after a federal judge ruled that its “supply-chain risk designation” was illegal retaliation.

2. Dr. Dre said he uses AI in his music production, comparing it to synthesizers and drum machines and calling critics “afraid of learning new things.”

3. NVIDIA reportedly agreed to buy Hugging Face for $12.9 billion, a deal that would give the chipmaker a major stake in the open-source AI ecosystem.

🤦‍♂️ AI fail of the week

Damn. You guys must’ve practiced that for ages!

The friction

Here’s a true story: Back in high school, I took a lot of notes.

Like, a shit-tonne of notes. (Oops! Excuse my language. I meant “shit-ton.”)

But when it was time to prep for my final exams, I flipped through my notebooks full of scribbles…and just ended up studying from textbooks instead.

It’s not that my notes sucked. It’s that pages of handwritten words aren’t exactly great for quickly grasping core concepts.

(Yes, the process of writing things down is valuable in its own right, but still.)

I’m sure I’m not alone in this.

We jot down handwritten lists, journal by hand, sketch stuff on the whiteboard during meetings, and leave all kinds of other written trails behind.

Then they often just…sit there.

Subscribe now

The fix

Luckily, there’s a simple way to turn all that text into a “big picture” overview.

Snap a photo (or a few) of your handwritten notes, feed it to a chatbot, and paste this:

“Distill this into a simple mind map that shows how the core ideas connect.”

This approach works with any multimodal model that has vision capabilities. (Nowadays, that’s most of them.)

I don’t have my high school notes handy (I’m old), but here are ChatGPT, Claude, and Gemini results for this Wikimedia handwritten page. (Click to view full image.)

ChatGPT and Gemini used image generation while Claude coded an SVG version, but they all managed to create order out of chaos.

Do this now

1. Snap a photo of a note you want to organize (or dig one out of your camera roll, you note-photographing maniac)

2. Drop it into your chatbot of choice with the prompt above

3. See if the mind map helps you make better sense of things

Tips for best results:

• Take the photo in good lighting conditions

• Use the best possible resolution (the more detail, the better AI can “see” it)

• Use the highest available “thinking” or “reasoning” effort

• If AI gives you a text output, add the word “visual” to your request

Go try it and let me know how it works for you!

Know someone who’d like this? Share it!

Share

Dig deeper [paid]

Notes-to-mind-maps are just one use case.

You can try this kind of AI voodoo wizardry with all kinds of pairings.

Read more

Welcome back to the weekly AI news roundup.

👉 See all 110+ paid subscriber bonuses

Here’s last week’s AI news roundup:

👩‍💻 AI releases

1. Alibaba launched HappyShrimp 1.0, an AI music model that can turn simple text prompts into full instrumental tracks or entire songs with lyrics. (Try for free.)

2. Anthropic news:

   1. Claude can now send emails via Gmail and manage your Google Drive directly from the online chat interface. (Paid plans only.)

   2. Computer Use, Skills API, and Files API are now available for developers to build with, so they can create more advanced and feature-rich apps.

3. DeepSeek launched V4-Flash-Vision-Exp, a multimodal model that adds image understanding while matching V4 Flash on text and agentic reasoning.

4. ElevenLabs rolled out Eleven v3 Conversational, a real-time speech model that supports 70+ languages and comes with a library of 11,000+ voices.

5. Google news:

   1. Gemini AI Pro is now free for a year for college students and comes with interactive visualizations, Deep Research tools, and study notebooks.

   2. Gemini in Chrome rolled out to Android users in the US with an Auto Browse agent that can perform tasks in the browser on your behalf.

6. Meta news:

   1. Meta AI Mac app can see your screen to answer questions about what you’re working on and also comes with Google Workspace tools.

   2. Pocket is an app that lets you create and share small games built from text prompts, now rolling out in the US. (Download for Android or iOS.)

7. OpenAI news:

   1. ChatGPT launched an Apple Messages plugin for Mac that lets you search conversations, catch up on threads, and draft replies using AI.

   2. ChatGPT for Teens comes with built-in safety protections and a Study Mode that uses guiding questions instead of direct answers. (My post on the topic.)

8. Ornith open-sourced Ornith-1.5, a family of self-improving models, with the flagship one allegedly matching the performance of Claude Opus 4.8.

9. Perplexity now lets you email Computer at computer@perplexity.com, so you can send tasks from your inbox for it to work on.

10. Replit launched Free Mode for Core subscribers, which lets them create 30x more content using GPT-5.6 Luna with limits that reset every five hours.

11. Runway released Ruby, a model that converts SDR video to 16-bit HDR in ProRes and EXR formats for professional editing workflows.

12. Salesforce launched Slack Code with dedicated channels where designers, developers, and PMs can collaborate directly with coding agents.

13. Stability AI upgraded Stable Audio 3.0 with a DAW plugin and richer web editor that lets musicians generate, tweak, mix, and extend tracks without restarting.

14. Tripo launched P2.0 Preview, a 3D model generator that creates game-ready assets from image or text prompts.

🔬 AI research

1. Cursor previewed its code-hosting platform Origin in early beta, with repos, pull requests, and GitHub sync built into the Cursor editor.

📖 AI resources

1. “2026 AI Marketing Industry Report” [REPORT]: a look at how marketers are adopting AI tools and their platform preferences by Social Media Examiner.

2. “How Much of the Internet Is Written With AI?” [STUDY]: interactive report by Pew Research Center tracing the spread of AI-generated content.

3. “Young Adults Increasingly Wary of AI” [SURVEY]: study of Americans’ attitudes toward and concerns about AI by the Pew Research Center.

🔀 AI random

1. Amazon has been bulk-buying rare books and scanning them for AI training data after destroying their bindings (404 Media investigation).

2. OpenAI paused training of its frontier model (codenamed Astra) after early tests suggested it could pose serious cybersecurity risks.

🤦‍♂️ AI fail of the week

Wait! Where did that coin go?! That went too fast for me!

I’m trying this new short-and-snappy post format with a problem + solution angle. Let me know what you think of it in the poll at the end.

TL;DR

The friction

“Prompt engineering” and other hacky prompt techniques are largely a thing of the past, but people still chase elaborate off-the-shelf prompts written by others.

You may think you don’t know how to write good AI instructions.

The thing is, you almost certainly already have.

I bet your chat history is full of conversations where AI eventually did what you needed…even if you had to yell at it a few times:

But did you know that every time you corrected the chatbot, gave it new input, or asked for a different approach, you were also writing your instructions in the process?

All that’s left is for you to turn those long chats into something you can reuse.

Subscribe now

The fix

Whenever you end a successful AI session, tell your AI chatbot or agent this:

• Use “skill” for agentic tools like Claude Code or ChatGPT Work (formerly “Codex”)

• Use “customizable instruction”1 for simple chatbots like ChatGPT, Gemini, etc.

The agent will simply build the reusable skill for you.

The chatbot will hand you an instruction with customizable [placeholders] that you can save for later. (You can also use it to create a custom GPT or something similar).

Do this now

1. Open your go-to AI chatbot or agent

2. Find a conversation that gave you the result you wanted

3. Paste in the above line

4. Profit

The best candidates are conversations about tasks that:

• You have worked on at least 3 times

• You expect to keep doing

• Are complex enough to call for detailed instructions

Go try it and let me know how it works for you!

Know someone who’d like this? Share it!

Share

Share your thoughts

I’d love to know your take on this kind of shorter “Quick Fix” post.

Welcome back to the weekly AI news roundup.

See all 100+ paid subscriber bonuses

In case you’d missed it, here’s this week’s Thursday post:

Here’s what happened in AI over the past two weeks:

👩‍💻 AI releases

1. Anthropic news:

   1. Claude Cowork can now join your Chrome session in a side panel, so you can sync sessions between web, desktop, and mobile on any device.

   2. Claude Tag now gets context from your entire Slack channel instead of just individual threads, making it better at knowing when to jump in.

2. DeepSeek launched V4-Pro, an upgraded agentic model with adjustable reasoning efforts that let you balance speed vs. depth depending on the task.

3. Google news:

   1. Gemini now connects to even more third-party apps like OpenTable, Pandora, Ticketmaster, and more, so you can access them without leaving the chat.

   2. Gemini 3.7 Flash is smarter at coding and tool use, handling complex workflows at half the price of its predecessor.

   3. Google Ads & Google Analytics have added AI agents that can manage your campaigns as well as new visual dashboards and benchmarking tools.

4. Meta open-sourced Muse Glimmer, an agentic model small enough to run on a consumer device for local coding and file management. (Try on Hugging Face.)

5. Microsoft launched MAI-Image-2.6, its strongest image model to date with improved portrait and 3D rendering skills. (Currently #2 on Text-To-Image Arena.)

6. OpenAI news:

   1. ChatGPT desktop app for Linux (now in preview) combines ChatGPT, Codex, and Work features in a single native workspace.

   2. Computer History for the ChatGPT Mac app can track your cross-app activity to build AI-accessible timelines and memories.

7. xAI news:

   1. Grok 4.6 is better at coding, research, agentic tasks, and visual work across complex projects, catching up with current frontier models.

   2. Grok Bot lets you deploy always-on AI agents that complete multi-step tasks inside your apps and websites around the clock.

8. Z.ai open-sourced GLM-5.3, a coding model with 50% better performance and stronger long-running agent workflows across complex projects.

🔬 AI research

1. Google introduced SL2T, a sign-language-to-text model that lets Deaf users sign to type, search, message, and interact with Gemini on Pixel 11.

2. Tencent previewed HunyuanWorldClaw, a system that generates explorable 3D worlds with editable terrain and objects from a single text prompt.

📖 AI resources

1. “From Assistance to Execution” [REPORT]: insights from several studies on how companies are transitioning from chatbots to agents by OpenAI.

🔀 AI random

1. Anthropic, Spotify, and Substack are all pushing for AI content transparency with disclosure tools, persona badges, watermarks, and more.

2. SpaceX completed its acquisition of Cursor AI, with the popular vibe-coding team joining SpaceXAI to build Grok coding products.

3. Twitch now shares your chats and streams with Amazon for AI training by default, so creators have to manually opt out in privacy settings.

🤦‍♂️ AI fail of the week

If you want to learn how to make pasta from scratch…maybe looks elsewhere?