Lately I’ve been noticing something fascinating: the rise (and maybe saturation) of vibe coding apps. Every other day it feels like a new one pops up — just recently I played around with CatDoes. Right now, there are roughly 30–40 vibe coding tools out there, and the number keeps climbing. We’ve got some big names already — v0, Lovable, Bolt.new, Replit, Base44 and plenty of others begging for attention.

What’s interesting is that each of these platforms comes with its own default UI style. Spend enough time browsing product landing pages and you can start to play “spot the tool.” Oh, this one’s v0. That one’s definitely Lovable.

I’ll give you an example. I was hanging out with some friends recently, and one of their phones started talking — kind of like Siri does randomly. I immediately knew that it was Google Gemini. That’s the same thing happening in design right now. Each vibe coding platform has its own recognizable “voice,” its own design aesthetic, and once you’ve seen enough of them, you start spotting the patterns everywhere.

And that fingerprint is starting to look a lot like an emerging AI design language.

Here are some dashboard UI examples from loveable.dev, bolt.new, vercel, next.js, shadcn.io While they look different you can start to see AI’s “taste” come through.

The new baseline for design

Here’s the thing: these AI-generated designs aren’t bad. In fact, they’ve already become the baseline for a huge number of new products. If tomorrow every healthcare or government website leveled up to this baseline, I’d be thrilled — because honestly, so many digital experiences still suck.

But a baseline is just that. When every landing page starts to look like a template, the hunger for creativity kicks in. This is where designers have a huge opportunity.

Why designers should be excited

If you’re a designer, you can use these tools to:

• Prototype faster.
• Play with visuals, animations, and micro-interactions you’ve always wanted to try.
• Customize and push beyond the generic defaults to stand out.

Even if you’re using the same vibe coding platform as everyone else, your design instincts give you an edge. Thoughtful tweaks and unique expression are what will separate you from the AI-baseline crowd.

One aspect I really love is the collaborative feel. Sometimes I’m working with AI almost like it’s a creative partner — bouncing around ideas, letting it surprise me, and following threads I wouldn’t have thought of on my own. Other times, I’m more of a context engineer — giving very strict instructions because I know exactly what I want. The fun part is flowing between those modes. It’s a new kind of creative rhythm, and it makes the work feel more alive. I’ve even launched my first SaaS products this way — built from the ground up through that back-and-forth collaboration.

I think people often assume AI is just a shortcut. And sure, it can save time, but when you use it well, it’s not about cutting corners. It’s about reshaping your process. Personally, I feel like I make more informed decisions now, not less.

For example, I don’t think I was wrong betting on my current tech stack, choice for my personal passion project, lil’ chef. I did a lot of research — with the help of AI — before committing to it. Because these newer AI-forward stacks are incredibly LLM-friendly, they’re easier to work with. They’re “LLM/Agent accessible. For example all of the documentation is AI first meaning it’s designed to be easily ingested by LLMs or agents.

Here’s an example of Vercel documentation with a copy markdown feature.

For me, this feels like a career unlock. I always wanted to code, but being dyslexic made it tough. I’d have ideas for visuals and interactions that I just couldn’t bring to life myself. Now I can. These tools have opened that door — and it’s fun. Honestly, I’ve never had this much fun and felt so refreshed as a designer.

I do want to acknowledge that I know my experience doesn’t reflect all designers. I know designers who are starting to feel burnt out or creatively discouraged. ❤

Final thoughts

I think we’re at a really exciting moment. Vibe coding in product design has set a new baseline, but the real magic is going to come from what designers do on top of it. It feels like we’re watching a new design language form in real time — one that’s both familiar and still wide open for interpretation.

And looking ahead, I think the same is true for the big players — OpenAI, Anthropic, Gemini, xAI, and whoever else enters the race. What’s going to set them apart isn’t just the raw model power; it’s going to be the experience. Just like I prefer certain LLMs for specific tasks because of how they feel to use, the user experience is going to become the real differentiator.

That’s my big thought, at least. I’d love to hear how you’re seeing it. Do you agree that AI is shaping its own design aesthetic? Are you excited about it, worried about it, or somewhere in between?

A shift is happening was originally published in Bootcamp on Medium, where people are continuing the conversation by highlighting and responding to this story.

An old draft I think still worth sharing. These thoughts are probably outdated even though this is only ~6 months old.

I want to just share where I’m at. Maybe it will resonate with you too. Is AI creative? Is AI taking my creativity away? Is what I produce with AI valuable? Is it ethical? What is art? Have we entered into a new era of art? Will we ever really know where the machine ends and the human begins?

Wednesday 5:59 PM — stream of thoughts

These and many more questions sometimes clutter my brain. It started when I dove head first into the early beta stages of Dall•E, Midjourney, Runway, and so many more. Things are changing. Things have already changed. It’s here and it’s moving so fast. I’m in it everyday and I find it hard to keep up.

But before we dive into all of that I think it’s important to introduce myself so you have a better understanding of my perspective.

Since I was young, I’ve been fascinated with technology. As soon as I could earn my own money I was spending it on the latest gadgets. When I was young, I just always remember my dad having really cool electronics around the house. He sold the first car cell phones and was what they call an early adopter. My mom was more on the art and creative side. When I was young, they ran and owned a photography studio. She opened a shop when I was in high school called Christy’s Creations. We had a kiln in the garage. I was always creating.

I remember getting on the internet for the first time, listening to the sound of the dial-up, and how it saved my appendix from rupturing thanks to my mom Asking Jeeves about my symptoms. I remember making my first designs in Paint and saving them to floppy disks. That love of technology has never faded. I guess, I’m also an early adopter. I think technology has the power to save the world and to destroy it.

Thursday 1:29PM

So, I identify as a curious creative. Anyone else? Really, this just means is I want to try everything. I’m a Product Designer by day, but I also want to be a 3D artist, an animator, a film producer, a psychologist, a game developer, a ceramist … you get it. I want to try it all. Generative AI has sparked something in me that I haven’t felt in a long time.

I’ve never felt more creative and creatively challenged than I have in this moment in time.

The definition of creativity is the capacity to come up with something new and innovative that has value to yourself or others. I’ve tried to bring this perspective to all of my AI interactions.

2:17

I do feel like I have a unique perspective since I’ve been immersed for so long. I’ve been able to see things progress and patterns form that others who joined in a little later might not have. If you’re newer here, then this next bit is for you.

It’s important to understand where we are. I’m sure at this point you have a good understanding of what generative AI is. It’s different from traditional AI because it can create entirely new content. That’s what feels like magic and gets everyone so excited. It’s also what’s freaking everybody out.

I am going to leave a lot of stuff out during this recap, but I’ll try my best to capture what I believe were some of the important moments in the last couple of years.

Things are moving so fast — so fast that nothing has ever moved this fast before. It’s‌ wild to look at the explosion in the curve and then realize that it was only a couple years ago. ChatGPT kicked off the big tech arena arms race and took the world by storm. It seemed (still seems) like every week there are major developments and breakthroughs in not just one but almost all sectors.

To give a better perspective on how quickly image models have progressed. These images were generated in Midjourney, showing versions from v1 to v6 (current is now v6.1) in just less than a year.

Outside of creativity, outside of coding and engineering, the progress we’re making in other areas like math and science is really incredible. For example, a year ago, GPT-4 passed the U.S. medical exam in December 2022, with a 60% accuracy rate, without previous reinforcement or training. It’s way more advanced now, outperforming the average human in social sciences and humanities subjects.

AI is also help discovering new classes of antibiotics, detecting wildfires, and even deciphering ancient scrolls. And it’s making us question our laws and ethics on a global scale.

There are obviously still a lot of ethical implications we haven’t solved yet. Last year, there was a rise in AI incidents and controversies, and it’s still skyrocketing.

Deepfakes have been around for a long time, but they’re now easier to create and more accessible. Recently, someone made a deepfake pretending to be the CFO of a company and got $25 million. Air Canada deployed a generative AI chatbot that started learning and providing refunds, which a judge decided they had to honor.

Does anybody remember any of these? Elliot Higgins generated images of Trump getting arrested in MidJourney in March of 2023. This caused MidJourney to cut their free tier. A fake image in a tweet caused the market to dump $500 billion in five minutes in May of 23. The viral image of the Pope in a puffer? I mean, that was just awesome.

Why were these images so important?

The images highlighted the potential dangers of AI-generated images and the need for fact-checking and media literacy. Experts warned that the technology is advancing rapidly and that it’s becoming increasingly difficult to distinguish between real and fake images. The incident also raised concerns about the potential misuse of synthetic media to spread disinformation and undermine trust in images

There was a song, “Heart on My Sleeve,” that used the likeness of The Weeknd and Drake and they weren’t happy, which caused the music industry to pay attention and start discussing and debating where this could go. On the flip side, Grimes came out said she wasn’t mad about it and offered to share royalties if people used her likeness to create something awesome. At this point, almost every creative sector was starting to pay attention.

Most disturbingly, a lot of AI-generated content is being used for porn and revenge porn. One of the recent victims was Taylor Swift. In response, her fans flooded the internet with the protective hashtag #ProtectTaylorSwift to counter the damaging images, showing how powerful a unified group of swifties (people) can be to step in when the laws and regulations aren’t sufficient.

So where does that leave us creatives?

We’re living in an era of collaborative tech, where we can become directors and technology acts as a collaborator. We won’t know where the machine begins and the human ends or vice versa.

Ten months ago, we were laughing at AI-generated videos like Will Smith eating spaghetti. Now, OpenAI launched Sora, China launched Kling, Luma launched Dream Machine and Runway launched Gen-3, the most advanced AI video generation we’ve seen, and the worst it will ever be. Before Generative capabilities updates and advancements like this would often take months, sometimes years. Each new version that we see is a massive leap and we’re at a point where we already can no longer determine what is AI and what is not.

14:18

We’re already living in an artificial reality, consuming content without knowing if AI was involved. In the future you can assume context windows will be large enough to handle massive datasets, making interactions more powerful. Images, video, audio, and text will become indistinguishable from human-created content, making it essential to have these conversations now.

Here’s where i’m at with my relationship with AI.

I’ve been living in the mind space of: If it’s imaginable, it’s achievable. AI has helped me unlock creative potential I didn’t know I could before. I’ve explored various creative projects like designing my first book by collaborating with AI tools like Midjourney (v4) and ChatGPT (3.5). And guess what! I used this book to propose! It’s already become a beautiful family heirloom. (read it here if you’d like) I also collaborated with it to design my fiance’s engagement ring. This was the moment I realized this was going to change the creative process tremendously. I’ve used it to create my first short film which was a finalist in Runway’s Gen-48 film festival. I’ve used to help me code games, websites, components, research and currently using it to build my first product!

Has creativity changed? I’m not sure but the process certainly has.

I’ve gotten to follow along with and connect with other amazing people in the AI community who are just doing some incredibly amazing things. I am so blown away by the amount of creativity I see. If anything it’s making me realize with the world at your fingertips your own creativity might be all that you have to set you apart.

18:52

My fiance and I used the help of AI for wedding planning. Helping in tasks like generating mood boards, organizing vendor quotes and honeymoon travel planning. AI has also been helpful in daily tasks like meal planning. We have our own custom ChefGPT with all of our dietary restrictions, likes, dislikes and a database of all our spices, kitchen appliances, etc. It even knows our comfort and skill level in the kitchen and our preference on a balance of meat vs veggie dishes per week. When I want to learn something new I can tell it to give me a more creative and advanced recipe because my goal is to experiment and learn. The personalized nature of AI makes it powerful in day-to-day life and work tasks.

AI has changed the creative process for many. One creative I follow got a deal with Cosmopolitan, creating the first AI-generated magazine cover. However, the misconception that it only took 20 seconds to make set false expectations for creatives using AI. The process involved hundreds of iterations, meetings, and traditional design skills.

30:46

AI has elevated me as a designer, allowing me to bring concepts to life and enhance my work. It keeps me creative, enabling me to explore and generate ideas that I find beautiful. AI tools like pen plotters have also sparked internal conversations about the collaboration between humans and AI.

35:41

Recently, I entered a competition called Gen 48 with a friend, using AI to create a film in 48 hours. This collaboration changed her creative process, making us realize the potential of AI to elevate our work. The experience showed that AI is not meant to replace us but to enhance and elevate our creativity.

So if I could give any advice it would be to learn as much as you possibly can not only from a productivity or creative aspect but more for so from the aspect that we all need to have like knowledge and expertise as to how these things work and what they're capable. It's going to shape the future and I believe that everybody should you know have a hand or have an opportunity to help shape that future.

I encourage you to use it even if you don't want to use it to create use it in another way use it for business stuff that you don't like to do use it when you're tired and don't know what to cook and wanna just take a picture of what's in your fridge and get a recipe like find creative ways in your lifestyle outside from the act of creating and you might just be surprised at what you find.

Conclusion

AI has been transformative, especially in learning and education. The personalized nature of AI has helped me learn more effectively and creatively. As we move forward, understanding AI’s capabilities and ethical implications is crucial, fostering a future where technology and human creativity coexist harmoniously.

This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories.

Subscribe to our newsletter and YouTube channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!

The Intersection of Creativity & AI: A Personal Journey was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

The world of product design is changing faster than ever, thanks to the rapid advancements in artificial intelligence. My journey into building with AI started modestly, experimenting with tools like ChatGPT and then expanding into more specialized platforms like Claude, Perplexity, and, most recently, artifacts. Each step has been a revelation, not just in capability but in the way these tools fundamentally transform how we approach design and prototyping.

The Evolution of AI in Design

It began with simple experiments — copy-pasting between ChatGPT and Visual Studio Code, running snippets in the terminal, and juggling dependency installations. I remember the excitement of creating my first custom game. Sure, it was just a Flappy Bird clone, but it featured my graphics, characters, and rules. Seeing it work locally felt like a superpower, even if the process was iterative and manual.

When Claude entered the picture, the game changed. Code generation became smarter, requiring fewer iterations to achieve the desired outcome. And then artifacts arrived, and that’s when it truly hit me: this changes everything. The ability to build through natural language — prompting rather than coding — opened new creative pathways.

Building Faster, Designing Better

For years, prototyping high-fidelity interactions or testing new component paradigms felt like bottlenecks in the design process. Tools like Figma and Framer are incredible, but they come with steep time investments. Translating an idea from your head into something tangible often meant spending hours perfecting animations or crafting detailed mockups.

Now, with AI, I can generate functional prototypes in minutes. A well-crafted prompt often delivers results that are “close enough” on the first attempt, letting me quickly iterate and refine. Seeing a concept in a working environment — not just a static prototype — reveals new possibilities and sparks immediate ideas for improvement.

Even more exciting is the ability to share these working prototypes directly with engineers. Instead of handing off a static design or a click-through Figma prototype, I can deliver something dynamic, something close to how it might operate in production. This shift bridges the gap between design and development, fundamentally altering how we collaborate.

The Designer-Engineer Hybrid

AI is pushing us toward a future where designers can become design engineers. Tools like artifacts don’t just speed up our workflow; they empower us to bring our ideas to life without waiting for someone else. For years, I felt blocked because I couldn’t code well enough to execute my visions. I’d have to hire or partner with an engineer, introducing delays and losing some of that initial creative spark.

But now, AI acts as a junior developer, enabling an iterative process where I can build, test, and refine in real time. It’s not just about speed — it’s about independence. The shift feels monumental. We’re no longer constrained by our technical skillset, and this democratization of building opens the door for designers to step into roles that merge creative vision with technical execution.

A Global Productivity Shift

The implications extend beyond individual workflows. As these AI tools become more accessible — free, even — they have the potential to spark a massive productivity boost across industries. Imagine the collective creativity of humanity, unleashed from technical or resource limitations.

When anyone with an idea can build without barriers, innovation accelerates. This democratization could lead to a renaissance of creativity, where people from all walks of life contribute to solving problems, designing better products, and imagining new futures.

Reimagining the Role of High-Fidelity Design

This evolution raises an important question: What does the future hold for tools like Figma? If AI can generate high-fidelity prototypes that operate almost like production code, will designers still invest hours in pixel-perfect and advanced prototyping features? I still think tools like Figma or other design tools will still be really valuable. A quick way to get a head start on your live prototype is often having a solid design as a base that a tool like cursor or Claude artifacts can work from. It also makes the prompt engineering a bit easier if you can speak better visually.

The answer might lie in how we define our roles. Instead of focusing on tools and workflows, designers can focus on vision, strategy, and problem-solving. High-fidelity design won’t disappear — it will transform. Prototyping in AI environments means iterating faster, collaborating more effectively, and delivering solutions that are closer to reality from the start.

Where We Go From Here

AI isn’t just a tool; it’s a collaborator. A really good one at that.

For designers, this means rethinking how we work, how we communicate, and what skills we prioritize. It’s a chance to help shape a future where the barriers between creativity and execution dissolve.

Remember, AI isn’t meant To replace you; it’s meant to elevate you.

Stay curious.

This article was made in collaboration with AI. For a full breakdown of my process don’t hesitate to reach out. I will publish the process in the coming weeks.

Thoughts on the future of product design in an AI first world was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

The AI podcast generator that is actually pretty darn cool.

Google’s NotebookLM is an innovative AI-powered research assistant designed to help users quickly understand, synthesize, and expand upon their source materials. This experimental product which is part of Google’s suite of experiments over at google.labs, has been making waves with its; recent feature updates. Mainly the new Audio Overview which generates hyper realistic ~5–20min podcasts based on your source material.

Sounds pretty cool right? NotebookLM has been around for a minute, initially unveiled as Project Tailwind during Google’s I/O conference in May 2023, NotebookLM has since been rebranded and made available to users aged 18 and above in the United States. As of September 2024, it has expanded to over 180 regions where the Gemini API is available. The best part? It’s free.

Key Features of NotebookLM

NotebookLM goes way beyond fun viral AI generated podcasts. Here are a few of it’s overall features you can benefit from.

• Source Integration: NotebookLM allows users to upload various document types, including Google Docs, Google Slides, PDFs, website URLs and now YouTube links. This versatility ensures that users can work with a wide range of sources, enabling analysis of video and audio content alongside traditional documents. Having access to your own or company google drive makes this a powerful tool for yourself and business.

• AI-Powered Analysis: The tool creates overviews of uploaded documents, providing summaries, key topics, and suggested questions to explore. This feature helps users grasp the essence of their sources quickly and efficiently.
• Intelligent Q&A: Users can ask questions about their sources, and NotebookLM provides answers with inline citations, linking directly to relevant sections of the original documents. This feature is particularly useful for fact-checking and deepening understanding.
• Note-Taking and Organization: NotebookLM enables users to create, save, and organize notes based on Notebooks. It can transform selected notes into outlines, tables of contents, study guides, or summaries, and now podcasts, making it a powerful tool for organizing and synthesizing information.
• Collaboration Features: Notebooks can be shared with others, facilitating collaborative research and learning. This feature has seen significant uptake, with users sharing notebooks and creating a network effect. (remember if you do this your data will no longer be private)

So basically a tool that let’s you learn in the style most beneficial to you. Who doesn’t want that?

The New Audio Overview Feature or AI Podcast Generator

The latest update to NotebookLM introduces the Audio Overview feature, which turns selected sources into engaging audio discussions. This feature is designed to help users who learn better by listening to explanations rather than reading them but also raises some ethical concerns. Here’s a quick breakdown of how it works:

• Initiation: To create an Audio Overview, users open an existing notebook, navigate to the Notebook guide, and click on the “generate” button.
• Processing: The system uses Google’s proprietary language model, Gemini 1.5 Pro, to analyze the uploaded sources and generate a dynamic, podcast-style discussion between two AI-generated hosts.
• Output: The generated audio discussion is produced as a WAV file, which can be downloaded and edited as needed.

Audio Overview Characteristics:

• Duration: The length of the audio discussions can range from a few minutes to approximately 20 minutes, with the duration seemingly determined randomly by the system.
• Content: The AI hosts summarize key topics from the source materials, using conversational speech patterns and sometimes metaphors to explain complicated concepts.
• Limitations: Audio Overviews are currently only available in English and may contain inaccuracies or audio glitches due to their AI-generated nature.

Sharing Audio Overviews:

• Public Sharing: Users can share Audio Overviews by clicking on the share icon in the audio player and generating a public share URL.
• Downloading: The audio file can also be downloaded and shared manually.

Integration with Other Tools:

• Google Labs Illuminate: An experimental technology that uses AI to adapt content to your learning preferences. This tool focuses more on a library of pre-existing research papers and books but you can still upload a URL to generate the audio overview. You can also see here they follow the same chip pattern from other projects like MusicLM to give a little more control over the output.

• Video Podcasts: Users can integrate NotebookLM with other AI tools like Hedra to create video podcasts by uploading the WAV file and adding visuals.
• Animated Transcripts: Tools like Descript can be used to create animated transcripts from the audio summaries, making them suitable for YouTube and other platforms.

Note that every output inside NotebookLM is grounded on the sources within those Notebooks.

Who is it for?

Well it’s for everybody. I think anyone can benefit from having access to a suite of tools like this especially since it’s FREE. Users have found various applications, including managing complex storylines for novelists and fan-fiction authors, and creating detailed descriptions of fantasy worlds for games like Dungeons & Dragons.

I think the primary target is students and educators, with preset functions to create study guides, FAQs and more. Imagine recording your lectures and being able to ask follow up questions, generate study guides, and now listen to it in an entertaining audio format. I wish I had access to this tool when I was in school.

Final thoughts and considerations

My initial reaction was “wow” — I couldn’t believe how realistic it sounded. My first generation was on an AI research paper that I was pretty familiar with. It actually helped me gain a new perspective I didn’t see before and develop an even deeper understanding of the topic. Powerful.

However as I’ve started to venture beyond academic papers and more into personal material and even something like the debate transcript (posted above), I’m starting to wonder about some of the ethical implications this might bring.

For example, I played some of these audio overview generations for family and friends and afterwards everyone was surprised to find out that it was AI generated. It makes me wonder, should there be a disclaimer at the beginning of each output? As more diverse voices start to get added will we be able to always recognize them as Google voices?

As I listened to the Audio Overview generated from my Creative Mornings talk on YouTube, I was initially thrilled to hear myself being discussed in a flattering light. However, as the episode progressed, I began to feel a growing sense of unease. The AI hosts were speaking as if they had attended my event, crafting opinions and insights based on their simulated experience, which struck me as both impressive and unsettling.

You can listen to the Audio Overview generation here. (the new sharing feature)

NotebookLM’s Audio Overview feature is a testament to the transformative potential of AI in reshaping how we consume, interact and learn with information. By turning complex documents into engaging audio discussions, it offers a unique blend of accessibility and entertainment that caters to diverse learning styles.

However, as we embrace these innovations, it’s important to remain mindful of the ethical considerations and potential implications. The realistic nature of AI-generated content raises questions about transparency and authenticity. As more voices and languages are added, ensuring that listeners can distinguish between AI and human-generated content becomes crucial. Perhaps implementing disclaimers at the beginning of each audio overview could be a step towards maintaining trust and clarity.

While these tools offer convenience and new perspectives, there’s a risk of oversimplification. The richness of long-form content and the critical thinking skills developed through traditional reading should not be overlooked. Balancing the use of AI tools with conventional methods will be key to fostering a well-rounded approach to learning. As we look to the future, NotebookLM stands as a powerful collaborator for students, educators, businesses and creatives. Its free availability democratizes access to advanced research tools. Yet, as with any technological advancement, responsible use and continuous reflection on its impact will ensure that we harness its full potential while safeguarding the integrity of our learning experiences.

What are your thoughts? How are you utilizing tools like NotebookLM?

This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories.

Subscribe to our newsletter and YouTube channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!

Everyone’s new favorite AI feature — NotebookLM’s new Audio Overview was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

Claude 3.5 Sonnet was officially released on June 20, 2024. When every other day it feels like there’s a breakthrough in multiple sectors in AI, it’s rare for a new release to really change up the landscape. Anthropic’s latest release, Claude 3.5 Sonnet, along with its innovative Artifacts feature, is doing just that. Here’s why this powerful combination is generating so much excitement in the tech community and why it’s a shift in how we interact with the UI of this technology.

It’s significantly better

Claude 3.5 Sonnet isn’t just an incremental update — it’s a quantum leap forward in AI capabilities. This model outperforms its predecessors and competitors across a wide range of benchmarks, including graduate-level reasoning, undergraduate-level knowledge, and coding proficiency. What’s even cooler is it achieves this while operating at twice the speed of the previous top model, Claude 3 Opus.

Visual Intelligence that Amazes

One of Sonnet’s standout features is its enhanced visual processing capabilities. It excels at tasks requiring visual reasoning, like interpreting complex charts and graphs. This makes it an invaluable tool for industries ranging from finance to healthcare, where the ability to quickly analyze visual data can lead to crucial insights.

The way that users can create these insights is what has everyone so excited.

The Game-Changing Artifacts Feature

Sonnet’s new capabilities are impressive, but it’s the new Artifacts feature that really makes it stand out. Artifacts allow users to interact with AI-generated content in a dedicated window alongside their conversation. This transforms Claude from a mere chatbot into a dynamic, collaborative workspace.

Imagine asking Claude to create a React app, and then being able to see, edit, and iterate on that code in real-time, all within the same interface. Or picture generating an interactive dashboard from a complex research paper, allowing you to explore the data visually or really however you want.

I use ChatGPT and Perplexity for coding. I’m a novice coder, but I’ve always wanted to be able to code. I often have all these ideas that require me to have coding knowledge so I can’t execute without help from an engineer. These tools have helped advance my skill set and I’ve learned a lot along the way. I’ve learned how to use the Terminal (though it still scares me) and Visual Studio Code. The thing is, with how these tools currently work, it’s a lot of back and forth. A lot of copy-and-pasting, prompting, and bug fixing.

With Artifacts I can go from idea to fully working interface in a fraction of the time. I’ve used it to make some fun html games; turn PDFs into interactive learning games; created a component generator, creative coding generator; a prompt generator for Midjourney and Runway’s Gen-3.

Recently, Anthropic announced the ability to publish and share Artifacts so other people can use and remix them. Remind you of anything? I wouldn’t be surprised if there’s a place soon to browse and use shared Artifacts.

You can check out the experiment image above here. I also share some cool examples from the community below.

AI for Everyone

I think the most exciting is how Claude 3.5 Sonnet is making cutting-edge AI accessible. It’s available for free on Claude.ai and the Claude iOS app. The caveat here is that while it’s free, you won’t be able to get very far. I am a subscriber and still hit the message limit. Claude recently announced Prompt Caching, which helps solve this issue but is only available for their API.

But I have a secret. You can access a similar experience for free at https://websim.ai/. Here is an experiment I did for a component generator.

The Future of Work and Creativity

The combination of Sonnet’s intelligence and the Artifacts feature opens up new possibilities for how we work and create. From rapid prototyping of software to generating interactive educational content, the potential applications are huge. As users explore these capabilities, we’re likely to see innovative use cases emerge that we haven’t even imagined yet.

A few awesome examples from the community

Allie K. Miller is one of my favorite AI leaders. She shows how powerful it can be to take data that you need to be able to understand and transform it to give you the information you need in whatever format you want.

People have been showcasing how easy it is to make games with Artifacts. Ammaar Reshi was able to recreate a 3D Doom came with sound effects and all.

@alexalbert__ on X made a piano app!

@rahul shared on X a react beats maker. Reminds me of the beat maker you can play with in the Tesla.

Conclusion

Claude 3.5 Sonnet and Artifacts represent a significant step forward in making AI not just more powerful, but more usable and integrated into our workflows. While it’s important to approach any new technology with a critical eye, the early results and user experiences suggest that this particular hype is well-deserved. As more people get their hands on these tools, we may be witnessing the dawn of a new era in human-AI collaboration with a new design pattern for GenAI interfaces.

Would love to know your thoughts and how you’re using Artifacts.

This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories.

Subscribe to our newsletter and YouTube channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!

Why Claude Sonnet 3.5 is worth the hype. was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

CNBC is introducing “Your Daily Olympic Recap on Peacock,” a personalized highlight reel narrated by an AI-generated version of legendary sports announcer Al Michaels’ voice.

When I first heard about this I thought, “wow, bold move.” The negative backlash companies have been facing when deciding to launch such bit AI features has been brutal. I’m sure you remember the latest Google AI search summary disaster.

This AI voice has been meticulously trained on AL Michaels’ past NBC appearances, capturing his distinctive cadence and delivery. It can even address viewers by name, adding a personal touch.

Take a quick listen here. I am not super familiar with the famous voice, but listening to the AI recap demo and Michael’s real voice I personally can’t tell the difference, can you?

Tailored Highlights for Every Fan

Starting July 27, 2024, Peacock subscribers can enjoy daily, personalized Olympic recaps. Imagine waking up to a curated playlist of the previous day’s top moments from your favorite events, all presented with the familiar voice of Al Michaels addressing you by name.

As a subscriber to Peacock you can enjoy:

• Customized 10-minute daily recaps
• AI-generated narration mimicking Al Michaels’ voice
• Personalized content based on user preferences
• Available on web browsers and the Peacock app

Subscribers can also tailor their recap experience by:

• Selecting up to three preferred sports categories
• Choosing up to two types of highlights (e.g., top competitions, viral moments)
• Inputting their name for personalized greetings

Personalization only AI can offer

This feature is capable of generating nearly 7 million permutations of personalized highlight videos. This allows for a highly customized experience tailored to users’ preferences, with each recap being approximately 10 minutes long.

This is something that would not be possible without the recent advancements in AI and takes personalization to a whole new level.

I was initially surprised that Al Michael’s agreed to the initial proposal. During an interview Michael’s admitted to not being very tech-savvy and didn’t fully understand the concept at first.

After hearing the AI version of his voice, Michaels described his reaction as “astonishing” and “amazing”. He told Vanity Fair,

“Frankly, it was astonishing. It was amazing”

After overcoming his initial skepticism and seeing the potential of the technology, Michaels ultimately decided to embrace the concept, stating,

“AI is here. Ready or not, here it comes, so you can either resist it or you can embrace it. I’ve chosen the latter.”

Technology Behind the AI Voice

The AI voice is created using generative AI and voice synthesis technology. It has been trained on Michaels’ past NBC appearances to capture his unique style, expertise, and elocution. The system can even address viewers by name, adding that personal touch to the experience.

The content is generated by pulling from NBCUniversal’s 5,000 hours of live coverage from the Paris Olympics to generate the highlights. Each recap is approximately 10 minutes long and includes highlights of the most relevant moments from the prior day for each viewer.

While I can’t find any information around what languages this feature will support it is safe to assume that if not now in the future we can imagine a voice like Al Michael’s being able to recap the Olympics in every language imaginable.

Keeping the Human in the Loop

While the use case is impressive, I think NBC is showing us that you can use this technology as a collaborative effort while keeping the human element, an important lesson. To ensure accuracy a team of NBC Sports editors will review all content including audio and clips before it’s made available to users, ensuring accuracy and quality control. To me this approach feels like the best of both worlds — leveraging AI’s capabilities to do something that’s never been done before while maintaining human oversight.

Broader Implications

I can’t help but think about the wider implications this could mean for sports broadcasting and creative industries as a whole. What does this mean for the future of authenticity? Will AI-generated content become the norm? How will this impact human commentators and content creators? How will we navigate the legal and ethical issues that rise? These are complex questions without easy answers. I’m excited to see how this experiment plays out during the Paris Olympics. It feels like we’re witnessing a pivotal moment in the evolution of sports media.

Looking Ahead

While I think I will always choose the authenticity of live, human commentary over AI, I’m cautiously optimistic about the potential of AI to enhance our viewing experiences. If it can help make the Olympics more accessible and engaging for a wider audience, that’s a win in my book.

As the 2024 Paris Games are kicking off, I’ll be eagerly watching (and listening) to see how this AI-powered coverage unfolds. Maybe I’ll find myself cheering for an AI-generated highlight reel just as enthusiastically as I do for the real time Al Michael’s coverage.

As this technology debuts during the Paris 2024 Olympics, its reception and performance will likely shape future applications of AI in sports media. Will this be a win for AI? What are your thoughts on AI in sports broadcasting? Are you excited about these developments, or do you have concerns? I’d love to hear your perspectives.

This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories.

Subscribe to our newsletter and YouTube channel to stay updated with the latest news and updates on generative AI. Let’s shape the future of AI together!

The Future of Sports Broadcasting: NBC’s AI-Powered Olympic Highlights was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

A great way to take your creativity to the next level when stacking generative AI tools to bring your ideas to life. Custom music and sound effects can take a piece to the next level. I this article i’ll share five AI Music Generation tools and how to start creating.

With a lot of generative AI tools you need to have an open mind. Not all tools are up to par with image generation outputs yet, but they will be soon. The earlier you get into using all these tools the better equipped you’ll be for when they’re producing amazing results and everybody wants in.

The other great thing is that these tools are still in their infant years; they are still in Beta, which means they’re free. This won’t be the case for long, especially with how fast things are moving. Let’s jump in!

Suno.ai

Suno AI is an AI system that can generate songs, instrumentals, and musical compositions based on text prompts.

Let’s start with Suno.ai that recently made waves in the generative AI community (even though it’s been around for awhile) due to the announcement of its’ integration with Microsoft’s Copilot. Suno started in 2022 and has made significant progress. When I first joined the Beta on Discord you could make short songs that were pretty terrible but very entertaining. Similar to those v1 image generations, if you remember those. Here’s two ways you can start generating your own fun tunes. 🎵

1. Microsoft Copilot

Step 1: Just head over to https://copilot.microsoft.com/ and either sign in or create your account.

Step 2: Over on the top right select Plugins and scroll until you see Suno and make sure it’s toggled to on.

Step 3: Start typing in the input field what type of song you want. You can even ring your own lyrics!

Here’s a link to the song I generated in the screenshot above. I’m at home with Covid as I write this, hence my song choice :P

2. Discord

Discord is where I first started interacting with Suno for their early Beta, and you can still generate songs in their server. Here’s how to-do it:

Step 1: Join their Discord server if you haven’t already.

Step 2: Over in the right side panel, click on the Suno Chirp Bot and send it a message.

Step 3: Use the /chirp command and press enter. A modal will appear. Enter the type of music. Your own lyrics or have ChatGPT write them for you. Voila! You’ve made your first AI generated Suno song in Discord.

Note They’ve recently limited users to 50 credits/day and each song is 10 credits.

Stable Audio

Similar to Stability AI’s image generation model Stable Diffusion, Stable Audio takes as input a user’s text prompt describing the desired output.

You’re not going to get your fun hip-hop lyrical jams here. Stable Audio can to text2audio and audio2audio. They have a pretty advanced Audio Diffusion Model that powers the generations.

I’m personally a really big fan of the UI in Stable Audio. It’s easy to navigate and pretty easy to use. You have quite a bit of control over your generations, but it may require some musical knowledge to get some incredible results. Things like defining your BPM, specific instruments, and so on. Here’s how to start generating:

Step 1: Sign up for an account at https://www.stableaudio.com/ and go through their nice onboarding. Note that while this is a free tool, you’re limited to 20 generations/mo.

Step 2: Write your prompt. Compared to Suno, I found prompting can feel a bit like being a composer, which I’m not. You have so much freedom with this tool, it can be hard to know where to start. If you don’t have much knowledge in music that’s okay, you can go with how you want to feel! Here are a few example prompts you can try out:

Winter afternoon, cup of tea, smooth vibes, slow tempo

A disco track that blends elements of Chicago and New York club scenes at 115 BPM. Use a driving drum machine, layered synthesizers, a prominent bassline, piano, and guitars to create an instrumental piece. The track should be clubby and euphoric, capturing the vibrant energy of disco with a modern twist. Focus on creating a groove that is both infectious and nostalgic.

Lofi hip-hop beat

An ambient techno track inspired by the tranquility of a Scandinavian forest. Utilize an 808 drum machine, with an emphasis on the 808 kick, claps, shakers, and synthesizers. Incorporate synth drones to create an ethereal, natural atmosphere. The track should be peaceful, beautiful, and instrumental, with a BPM of 122, reflecting the serene and untouched beauty of nature.

Remember it can make sound effects too!

If you want to just jump in and not worry about prompting, you can just grab something from their prompt library which offers a large list of great starter prompts and a great way to learn.

Be sure to play around with all of the other features too, like:

Input audio

Select one of your generated tracks to be used as input audio. This will populate the input audio area with your selected track.

Input strength

This slider controls how much of your selected audio is imposed on the final result.

Higher percentages means that less diffusion will be done on your audio, whereas lower percentages allow for more diffusion and thus more variation.

Steps

Informs the amount of generation steps used to create your audio track. Higher step count = greater processing. This can increase the quality of your audio slightly. It’s set to 50, which is said to be the sweet spot.

Number of results (paid only)

This controls the number of generated tracks returned while generating. Only accessible on the pro plan — with a maximum number of five generations at a time.

Seed

By default, this input is set to ‘random’. By adding your own seed, this informs the specific arrangement of noise used to generate your audio. When using a model that supports deterministic results, providing a seed alongside the other settings used to generate will result in that same audio being generated again.

Note: when generating multiple results, remember to have the seed set to ‘random’.

Prompt strength

Controls how closely the model attempts to guide the audio to your text prompt.

Happy generating! Onto the next tool. 🎵

Google’s MusicFX

MusicFX is an experimental technology that allows you to generate your own music. MusicFX is powered by Google’s MusicLM and uses Google DeepMind’s novel watermarking technology, SynthID to embed a digital watermark in the outputs.

This interface got a major lift recently. When the Beta first came out, it was called MusicLM and it looked like this:

The generations were decent but nothing compared to the outputs now. Super impressive!

Step 1: Head over to the Google’s test kitchen and log in with your Google account.

Step 2: Start generating! I am a HUGE fan of the new design, especially for prompting. When you start writing your prompt it will turn certain words into what they call Chips. Chips help you explore other ideas by changing parts of your prompt while your original prompt is still being used. It basically pulls similar words and surfaces those as easy pick lists or sliders in the input field.

There are some other neat features under Settings.

You can generate up to 70 seconds, and if you turn on Looping, you’ll have an endless melody. Here’s the one I generated:

The great thing about MusicFX is that it’s unlimited generations and free. The downfall … it doesn’t save ‌generations. Meaning once you log out and log back in, it’s a blank slate. Now onto the next! 🎵

P.S. Check out the Instrument Playground too.

Riffusion

Riffusion brings your typed lyrics to life. It operates by transforming the prompts or inputs from users into spectrograms, which are essentially visual depictions of sound. This tool then skillfully converts these visual representations back into audible sound, allowing users to listen and fine-tune the audio as per their preferences.

Similar to Suno, Riffusion lets you create lyrical masterpieces. Well, maybe no masterpieces, but you get the point. It also generates an image to go along with the type of music you select, which is fun. Here’s how to start generating:

Step 1: Head over to https://www.riffusion.com/ and create an account. Once you’re in, you can explore trending Riffs or go straight to Create.

Step 2: Click into the input field to open the popup. You have two options here. Lyrics and Describe the sound. You can use their generative feature to write lyrics for you, go to ChatGPT, write your own or even record them.

The describe section is how you want your music to sound or feel.

Tip: Try leaving the lyrics empty and just describe the sound to get music with no lyrics

Press Riff and wait for your music to be generated!

You’ll get three different riffs as a result. From here you can keep generating or click on the more to share your results, get the mp3 or pull out specific pieces.

Overall, Riffusion is super fun and more importantly FREE. The voice generation is behind Suno and others but I bet it won’t be long until it’s caught up.

Meta’s open source MusicGen

Meta has a ton of amazing open source models that they have been contributing, and one of those is MusicGen, a text2music generator.

MusicGen (AudioCraft) is a single-stage auto-regressive Transformer model trained over a 32kHz EnCodec tokenizer with 4 codebooks sampled at 50 Hz. While Meta doesn’t have a fancy UI to interact with this model there are a few options.

Option 1: Replicate

Replicate lets you run and fine-tune open-source models. Deploy custom models at scale. All with one line of code. If you’re not familiar with it, I highly recommend it. You don’t need to have an account to start generating music.

Step 1: Head over to this URL. You’ll see the interface above. While this can seem a bit overwhelming, you can just focus on the prompt! Let’s reuse a prompt from earlier and leave everything else default:

Winter afternoon, cup of tea, smooth vibes, slow tempo

Here’s the output:

Note that while this is free, it does still require GPU usage, so you’ll reach a “time limit” which will refresh after awhile.

Option 2: Huggingface

Similar to Replicate, Huggingface is an open-source community that offers tools for building, training, and deploying ML models. Hugging Face also provides the infrastructure to run, demo, and deploy artificial intelligence (AI) in live applications.

Step 1: Go to this huggingface space, and type your prompt! While this interface is a lot simpler, it’s a lot more limited than trying it out on Replicate. You’ll get a 15-second output, but you can upload a file or use your mic to influence the generation. Very similar to a reference image in an image generator.

Overall, this shows exciting promise for advancements in the Generative Music scene. I imagine the sound designer who now has access to these tools to create custom sound effects, or the musician who can use these tools to design and build songs in the ideation phase. What are your thoughts on generative music?

This story is published on Generative AI. Connect with us on LinkedIn and follow Zeniteq to stay in the loop with the latest AI stories. Let’s shape the future of AI together!

5 free ways to generate music with AI was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.

The dawn of Artificial Intelligence (AI) has not only beckoned a slew of technical advancements but has also initiated a profound discourse in the creative realm. Much like the ongoing ethical question of, “Is AI Art?”, the design community is now grappling with a similar existential question: “Is AI Design?” This inquiry transcends the narrow corridors of design and permeates other creative fields like photography, video, film and much more. The AI wave is not merely a transient phase but a new era in the evolving narrative of creativity, as it seamlessly melds with the tooling technologies creatives use everyday.

At the heart of this discourse lies a dilemma — to embrace the AI-driven tools that promise unprecedented efficiency and innovation, or to uphold the traditional ethos of design that values human intuition and manual skill. This is not merely a philosophical contemplation but a pragmatic concern for professionals. The reluctance to integrate AI in design practices might stem from a deeply-rooted ethical stance. However, as the hands of time endlessly march forward, those embracing AI are not only keeping pace but are pioneering a transformation in the creative landscape.

Generative AI has been a notable catalyst in this transformation, especially during the ideation phase. It doesn’t merely augment the output but stands as a collaborative companion in the creative process, offering an abundance of perspectives that might have taken a human mind much longer to conceive. The rapid transition from ideation to realization, facilitated by AI, is not just about efficiency; it’s about transcending the conventional boundaries of what design can achieve.

Imagine a scenario where a designer, traditionally reliant on prototyping tools and human only ideation, they now have the capacity to ideate much quicker and code their ideas into prototypes, courtesy of AI-collaborative learning and code-generation tools. This evolution is not just a leap in efficiency but a stride towards a domain where ideas are brought to life with a finesse and speed that was not possible before. You are now liberated from the constraints of your existing knowledge and tools, and poised for greater success in your field.

Maybe your a film producer or storyteller and you can now storyboard in less than half the time and create a conceptual video before moving into production. Here’s an example below where cardboard scenes were used to generate camera movement, set design and more using RunwayML GEN-1 (a video2video AI generation tool).

Abel Art on Twitter: "📦📦📦I am telling you that CP 👀 (Cardboard Production) is a possible alternate future of AI film making. Here again with @runwayml GEN-1 and this time a photo found on internet.This means:Total control over the set, the set design, and the props. No compromise.Cost... pic.twitter.com/Frk3cwutux / Twitter"

📦📦📦I am telling you that CP 👀 (Cardboard Production) is a possible alternate future of AI film making. Here again with @runwayml GEN-1 and this time a photo found on internet.This means:Total control over the set, the set design, and the props. No compromise.Cost... pic.twitter.com/Frk3cwutux

Perhaps you’re a product photographer seeking inspiration for your photoshoots before the setup. Alternatively, you might opt for crafting synthetic mockups and hard-to-obtain landscapes instead of using a camera, all for cost-effective results.

As we venture deeper into this new era of AI and creativity intertwined, the narrative is likely to evolve from “Is AI Design?” to “AI is a tool that elevates design and creatives.” However, amidst this transformative journey, it’s imperative to harbor a sense of inclusivity towards those who choose to remain anchored to the traditional practices of design. Their choice to uphold the ‘old ways’ is a valuable endeavor to preserve the legacy of our craft. In the future this might be a highly desired skillset. Similar to those who have preserved processes such as film photography.

The juncture we find ourselves in is indeed a historic marker in humanity’s timeline. The decisions made today will echo through the archives of history. It’s a time that urges not only acceptance of the new but a respectful acknowledgment of the old. There’s an inherent beauty in the preservation of our past, as it serves as a cherished reminder of our roots, even as we venture into uncharted territories.

The dilemma non adopters will face is that this new collaborative process between human and machine is making its way into the everyday tools that we use to create. For those who do not adopt and learn will fall behind from a tooling perspective but will be forced to embrace a symbiotic relationship between human and machine. Companies like Adobe are creating features into the creative process that feel natural and more as a companion vs removing creative autonomy. Putting you in the directors seat if you choose to take on that role.

Check out some of the features from Adobe’s Express announced at Max this year:

Watch how Figma sees the future of AI and Design:

The future might host a harmonious coexistence of AI-collaborative design and traditional methodologies. This dichotomy will not only be a testament to the evolutionary narrative of design but also a reflection of our adaptive and inclusive spirit. As we navigate through this metamorphosis, the hope is to foster a community that is not only forward-looking but also deeply reverent of its rich heritage. The dialogue between the old and new is a narrative of growth, respect, and an unyielding quest for creative exploration, with AI being a significant yet not solitary player in this grand narrative.

Have you been using AI in your current processes? Would love to hear your thoughts in the comments.

As always, stay curious.

This story is published on Generative AI. Connect with us on LinkedIn to get the latest AI stories and insights right in your feed. Let’s shape the future of AI together!

Is AI Design? was originally published in Generative AI on Medium, where people are continuing the conversation by highlighting and responding to this story.