I'm excited to invite you to our upcoming hands-on workshop where you'll learn to build your own AI agent using n8n and Telegram - no coding required!

Led by Amir Elimelech, Chief Customer Officer at UIDB, this 1-hour session will walk you through creating, deploying, and customizing a functional AI assistant.

Date & Time:

• 18:00 Israel (IST)

• 17:00 Berlin (CET)

• 16:00 London (GMT)

• 11:00 New York (EST)

• 08:00 San Francisco (PST)

What you'll learn:

• Building a working Telegram AI bot

• Step-by-step n8n workflow creation

• Customization techniques and best practices

• Practical implementation strategies

You'll leave with a functional AI bot, workflow templates, and implementation guides.

Perfect for product managers, content designers, no-code enthusiasts, and anyone interested in practical AI implementation.

Looking forward to seeing you there!

Sign up for free here

Hey 👋

I'm Yuval, the founder of one of the world's leading UX writing courses. I've built and scaled my business using AI and no-code automation, from helping hundreds of students to creating emergency response systems.

I'm inviting you to join AI Makers Hub, where I'll show you exactly how to implement AI solutions without coding.

You'll get:

• Free access to my courses and blueprints

• Real examples from my client work (like my AI image generator for a major soft drink brand)

• Weekly live sessions (at 50 members)

Free until we hit 100 members - join now!

Join AI Makers Hub

Let's build something amazing, Yuval

Discover how to harness the power of AI and APIs without writing a single line of code! In this tutorial, we'll show you how to create stunning video applications using popular no-code tools and AI services.

Learn how to leverage Claude AI to generate API calls effortlessly Integrate Luma Labs' video AI capabilities into your projects Use Airtable as a flexible backend for your no-code app

See the full video here

It all started when I stumbled upon an incredible voice cloning tool at a conference. The potential blew my mind, and I knew I had to try it out. So, I decided to take it a step further and create an AI agent that could book meetings for me using my own cloned voice. Here's how it all went down.

Step 1: Cloning My Voice

The first thing I did was clone my own voice using the tool I found at the conference. It was a surreal experience hearing my digital doppelganger speak for the first time. If you're interested in trying this yourself, there are several voice cloning technologies out there. Just make sure to do your research and choose a reputable one.

Step 2: Creating the AI Agent

With my voice cloned, I moved on to creating an AI agent that could handle booking meetings. This part was pretty complex, involving natural language processing and machine learning. I had to train the AI to understand and respond to requests related to scheduling and calendar management. It took some time, but the results were worth it.

Step 3: Giving My Agent an Avatar and Name

To make my AI assistant feel more personal, I gave it an avatar and a name. I found this step surprisingly fun – it was like creating a digital alter ego of myself!

Step 4: Connecting the Agent to My Calendar

For my AI to actually book meetings, I needed to connect it to my calendar. I integrated it with my existing calendar system, which allowed the assistant to check my availability and schedule meetings accordingly.

Step 5: Testing It Out

Now came the moment of truth – testing my new AI assistant. I decided to start simple and asked it to book a meeting with myself. I have to admit, I was a bit nervous, but also excited to see how it would perform.

The Results

I'm happy to report that the results were pretty impressive! Here's what I found:

• The AI agent was able to book meetings without any problems. It was quite surreal hearing my own voice confirm appointments!

• The whole experience was, in my own words, "quite wild." I found myself genuinely impressed by what the technology could do.

• Overall, I'd describe the AI assistant as "pretty cool." It's not perfect, but the potential is definitely there.

Some Hiccups Along the Way

Of course, not everything went smoothly. There were a few challenges:

• The agent was still a bit buggy and I had to do some editing to get it to work properly.

• It couldn't send emails directly, which meant I still had to step in for some tasks.

Final Thoughts

Creating my own AI assistant with voice cloning technology was an incredible experience. It's amazing to think about how this technology could streamline our daily lives and boost productivity.

However, I want to emphasize the importance of using this technology responsibly. There are certainly ethical considerations to keep in mind, especially when it comes to voice cloning.

Have any of you tried something similar? Or do you have ideas on how to take this concept even further? I'd love to hear your thoughts and experiences in the comments below!

Stay curious and keep innovating!

About seven years ago, I started building a community of UX writers.

As the community grew, businesses emerged from it like online courses and training programs, an agency, and more—all related to UX writing, of course. And these businesses are still going strong today.

Yet few realize that a huge part of their success was the use of various home-grown technological solutions. Even though I never learned to code, I realized that with a bit of effort, I could find the “lazy person’s” solution and do things myself.

I used pre-built solutions like Kajabi to host courses, while no- and low-code tools like Airtable helped me build databases and more. Finally, tools like Make became the glue that held it all together.

To this day, these solutions power our business—no outside assistance required. 

AI Makers Lab with Yuval

I’ve been teaching UX for years now, but recently I’ve been drawn to sharing my no-code knowledge with the world as well. Just like with UXW, I hope to build a community of “makers” who will lead the field in building automations and no-code software solutions, combined with AI, of course.

So I’ve done what everyone does when they want to share something with the world. Long story short, I’m now a YouTuber. 😅

In my channel, I’ll be teaching everything, and I mean everything I know about this field. For example, I’m super excited to share with you the latest creation from the AI Maker’s Lab. It’s a web app for e-commerce companies that generates custom AI images for their products. In fact, the app has already earned two paying customers in its first week!

It's crazy how quickly you can generate viable digital products when you know how to combine no-code with AI tools—products that actually make you money! We’re facing a future when any random teenager will be able to launch the next big startup from the comfort of their parent’s living room. 

Small businesses will be able to scale and compete with larger enterprises. Your neighborhood barber will be able to build his own CRM full of business automations and applications, and maybe even an accompanying training program. Because hey, why not?

So let me show you what I’m talking about with our e-commerce AI image generator.

Endless custom e-commerce images

That’s the idea here: any e-commerce site will be able to generate images of their product in any background, context, or scenario one could think of. So how’d we do it?

Based on this video:

Step 1: AI model training made simple

First we have to train the AI on the physical product: remember, we’re talking about e-commerce here. 

For this we used Replicate and Hugging. These platforms make it possible for anyone to fine-tune powerful AI models without deep machine-learning knowledge. With a user-friendly interface, we can upload product data and train models that understand specific brand aesthetics and product styles.

2. Data management: no SQL required

Airtable is our secret weapon for managing complex datasets. Its spreadsheet-like interface is familiar to most, but it packs the power of a database. We've created a system to track product catalogs, training data, and model performance—all without writing complex queries.

3. Workflow automation: the magic of make.com

Here's where it all comes together. Make.com (formerly Integromat) allows us to create complex automation workflows visually. We've built a system that automatically:

• Syncs product data from e-commerce platforms to our Airtable database

• Triggers new AI model training when sufficient new data is available

• Monitors model performance and alerts us when optimization is needed

• Generates and sends performance reports to clients

The results look like this:

If you want a complete walk-through, check out the full video. 

Or, learn more about the service if you have an e-commerce product you’d like to generate images for.

The revolution will … be on YouTube

This project was just one of many to come and I’ll be sharing more on the AI Maker’s Lab channel.

What I love about this approach to building software solutions is that it’s available to all. You don’t need a degree in computer science or even to be able to write a single line of code. It helps level the playing field between gigantic enterprise companies with capital and resources and SMB with nothing more than an idea and a vision. 

I hope you’ll join the revolution and start “making” great things with these tools and techniques because the future of tech isn't just for coders anymore. It's for creators, problem-solvers, or anyone with a vision and the drive to bring it to fruition. 

So, are you ready to become a no-code AI maker?

Drop a comment below or in the full video and share your thoughts and ideas. Let's push the boundaries of what's possible with no-code AI.

LLMs like ChatGPT have opinions about seemingly everything.

But if you’ve ever wondered, “What do they think about me?” then I have good news and bad news: you can now find out, but you will be roasted mercilessly.

Meet Wordware and its Twitter Personality roast feature, an AI-powered tool that’s taken the internet by storm with its brutally honest roasts. It scans your X/Twitter profile including user info and tweets and then does some AI magic to analyze your personality, exploiting your weaknesses and saving them for roast fodder. 

And it’s remarkably effective. Check out the savage takedown it gave me:

“Ah, the self-proclaimed UX messiah of Tel Aviv! Your tweets read like a desperate attempt to stay relevant in a field that's moving faster than your ability to create another bootcamp. You're so busy trying to predict the future of UX, you've forgotten to actually design anything useful. Your obsession with AI is cute, but let's be real - even ChatGPT writes more engaging content than you do. You're like the guy who shows up to a party talking about NFTs when everyone else has moved on. But hey, at least you've got that podcast where you can ramble on about 'the future of tech' to an audience of, what, your mom and three bots? Keep chasing those trends, Yuval. Maybe one day you'll actually catch up to one.”

Ouch, that was brutal. But also, wow! That was amazing! 

This app pulls no punches and hits surprisingly close to home. But the joke’s on you AI roast machine—my mom doesn't even listen to my podcast. 🥲

What’s under the hood?

So how does it work? Since the entire app is open-source, it’s pretty easy to reverse-engineer it and find out. There are four major components.

1. Data scraping

Obviously, the first step is to scrape the data from the profile. For this, the creators used Apify, a powerful web scraping tool. 

It scrapes both the user’s profile info and recent tweets. Like a ruthless predator, studying its prey… 😑

2. Specialized prompting strategy

Here’s the secret sauce—carefully crafted prompts. The app uses a comprehensive template that guides the AI in generating personalized analyses. The prompt covers various aspects, including:

• Strengths and weaknesses

• Personality traits

• Career insights

• Quirky elements like "past life" suggestions and spirit animal comparisons

Here's an excerpt from the actual prompt used for the roast feature:

```

Roast. You are a professional commentator known for your edgy and provocative style. Your task is to look at people's tweets and rate their personalities based on that. Be edgy and provocative, be mean a little. Don't be cringy.

```

IMO, this is what makes this feature work. There’s an overload of crappy, mind-numbingly boring AI-generated written content out there these days. So to generate quality AI content—texts that people actually want to read—you need to work on your prompt game.

3. AI model integration

With the scraped data in one hand and soul-crushing prompts in the other, Wordware is then ready to bring the pain by feeding it all into an advanced LLM. I couldn’t find the exact model in the source code, but our friendly neighborhood Claude 3.5 Sonnet seems likely—it’s pretty popular right now.

4. Web presentation

The final piece of the puzzle is a slick website with which to deliver the smackdown. It displays the results and encourages sharing, which was a crucial component of its viral spread.

More than just roasting

The funny thing is, Wordware’s roast feature is just one small part of the larger product. The full Wordware app is designed to enable “anyone to develop, iterate and deploy useful AI Agents.” (So says its tagline.)

But going viral surely helped boost the entire product. In fact, its popularity skyrocketed so much that it broke Product Hunt.

The Price of AI popularity

While Wordware's success is impressive, it raises a few questions about the cost of running such AI-intensive applications. Each personalized roast requires a separate AI generation, and with millions of users, those compute costs add up fast. 

It's a reminder that viral AI products need solid monetization strategies to remain sustainable. But it looks like Wordware is doing OK so far.

The no-code revolution

Wordware is part of a growing trend of no-code and low-code AI startups. These tools are democratizing access to powerful AI capabilities, allowing entrepreneurs to quickly build and launch products without deep technical expertise. However, as Wordware shows, scaling these products successfully requires careful planning and resource management.

What's next for AI roasts… and the future of mankind? 

As AI language models continue to improve, we can expect even more creative and personalized applications. Which should be fun.

But will something like Wordware’s roast feature eventually transcend the boundaries of “The App Formerly Known as Twitter” to roast our entire digital lives? Will we all pretend to love it while secretly whimpering on the inside like lost puppies? 

Only time will tell, but one thing's for sure—the internet's appetite for quality AI-generated content shows no signs of slowing down.

Happy roasting, and share yours!

When ChatGPT first burst onto the scene, Google went into full-on panic mode. We’re talking hair-on-fire, layoff-hundreds-of-employees panic.

Why?

Google saw ChatGPT as a threat to its search engine business—Google’s bread and butter.

But once the dust settled, it became clear that ChatGPT and its ilk filled a different need and niche from search engines. In part because they can’t be trusted to give factual answers. ChatGPT hallucinates too much. Too often, it’s a “liar, liar, pants on fire.”

Sure, ChatGPT can help with a high school English paper, but less so if you’re doing serious academic research.

But what if AI could access all of the internet and cite multiple sources for the information it’s presenting?

Thus thought former Google and OpenAI researcher Aravind Srinivas, who founded Perplexity AI to do just that.

Search vs. answer

Perplexity bills itself as an “answer engine,” or even a “knowledge discovery engine,” rather than a search engine. Its tagline, “Where knowledge begins,” hints at Srinivas’ philosophy that there’s no end to knowledge—it can always expand and grow.

Perplexity isn’t trying to replace Google Search or “play Google at their own game.” Rather, the idea is to rethink the standard search engine UI, replacing a list of links with factual answers to questions and relevant follow-up questions to help keep the knowledge flowing and expanding.

And while Srinivas recognized that Perplexity may still hallucinate on some queries, the gamble is that as the technology improves and gets cheaper and as hallucinations are reduced, they’ll be able to do what Google can’t or doesn’t want to do—provide accurate answers without a list of links.

The GenAI arms race

This idea of supporting generative AI query responses with links to the source isn’t unique to Perplexity. In fact, Microsoft’s Copilot has had links since its launch in 2023, and now even ChatGPT adds links to sources on some responses.

But Perplexity still has the upper hand in that it prioritizes trustworthiness by sourcing its information from selected reputable sources instead of weighing equally any and all results from a massive index.

This makes it the best choice for online research.

AI research

Maybe it’s because its founders were inspired by the academic idea that everything must be backed by a source, but Perplexity is the #1 AI tool for research.

Its ability to aggregate and synthesize information from various trustworthy sources makes it a superior tool compared to other GenAI tools. 

The use cases are vast, for example:

• Competitor analysis: getting detailed information about competitors, including recent product launches and market strategies.

• Academic research: Students and researchers can quickly find background information, relevant papers, and key concept summaries using Perplexity for their studies.

• Technology forecasting: Tech enthusiasts can use it to stay informed about emerging technologies, their impacts, and adoption trends.

• Cryptocurrency trends: Investors can keep up with the latest crypto trends, including emerging coins and regulatory changes, with concise market summaries.

• Real estate insights: Perplexity can offer up-to-date information on markets worldwide. Users can quickly learn about property laws, popular areas for investment, and current market trends.

• Finding coupons: Because Perplexity can access newer information, it can even retrieve functional coupon codes for seemingly any website. (link to vid)

Perplexity fetches info that’s so up-to-date that it even gave me an accurate (mostly) Olympic medal count:

Compare that to ChatGPT’s attempt:

You can see that ChatGPT is trying to support its data with a link to a source. But the source isn’t so good and it’s not able to make sense of the data the same way the Perplexity can.

It’s hallucinating so badly that it has Russia in 3rd place in the medal count even though Russia isn’t participating in the games.

And yet, Perplexity still has one mistake—an extra bronze medal for the USA in the table, even though the textual data below the table is correct. 

For this kind of info, Google Search is still best. But it does demonstrate how current Perplexity’s data is. 

Getting in focus

Perplexity has a user-friendly and intuitive UI. There’s a great feature called Focus that lets you narrow your results. It’s pretty self-explanatory.

Depending on what you’re looking for, there are times when it seems that Perplexity feels like a Google Search/ChatGPT hybrid.

And yet, even Perplexity still isn’t a serious threat to Google search … not yet anyway. 

Moreover, Google's business model extends far beyond search, with the tech giant generating substantial revenue from its cloud services, YouTube, and other ventures, providing a robust financial cushion against potential search competitors.

The hype and the heat

Perplexity's rise hasn't been without controversy. 

For example, there are accusations that the platform ignores robots.txt files, which is supposed to allow website owners to specify which parts of their sites should not be crawled or indexed.

This practice has allowed Perplexity to access and summarize content from behind paywalls, raising serious questions about copyright infringement and fair use.

When confronted with these allegations, Perplexity claimed that the company wasn't directly ignoring robots.txt, but rather using third-party scrapers that did so, failing to address the ethical implications of knowingly benefiting from such practices.

Final thoughts

As it currently stands, Perplexity’s value lies in its ability to combine advanced GenAI with real-time information retrieval, providing users with concise, accurate answers backed by reliable citations, making research efficient and accessible.

Perplexity has seemingly endless use cases. Give it a try and let me know if you discover any new and interesting ways to use it.

Fire, farming, steel… the iPod. Humans have invented some cool stuff over the years.

And every now and then, a new technology pops up and changes everything—a civilization-wide paradigm shift that alters the course of human history, if you want to be dramatic about it. 

Ships crossed the oceans, connecting peoples from worlds away. But these were soon replaced by airplanes, deepening our global interconnection.

Or consider how In the span of roughly 100 years, we went from music being a strictly live event, to the phonograph, cassette tapes, CDs, MP3s, and now streaming, giving us instant access to nearly all the music that’s ever been recorded.

And of course, the revolutions of cloud storage (buh-bye save icon) or the smartphone, which connected us with anyone, anywhere in the world at any time, and put the sum total of human knowledge in the palm of our hands.

These are not mere technological changes—they fundamentally altered how we interact with the world and with each other.

As UX professionals, we have a duty to keep up with these changes to help us deliver the best possible experiences to our users. 

So as we stand at what appears to be the precipice of another such upheaval—the AI revolution—what are some of the technologies that we can incorporate into our UX practice to help us ride the wave and not just get swept away by it?

I propose that simply learning to use generative AI technologies like ChatGPT isn’t enough. 

UX professionals should take it one step further and learn what's possible when we incorporate more advanced AI technologies. 

For example, LangChain.

What the heck is a LangChain?

LangChain is an open-source framework created by Harrison Chase in 2022—and I realize that may mean basically nothing to some of you. Fair enough. 

So how about this: LangChain is software that makes it easier to build apps that use large language models (LLMs) like GPT-4. Let’s say you want to build an app that leverages ChatGPT to help users blocking the internet for your kids id their laundry basket is filled. LangChain makes it easier to do that.

Now, for us UXers, this doesn’t mean we need to become developers. But understanding a tool like LangChain will help us understand what kind of digital experiences it’s now possible to give our users. And that is a very important thing, especially if you want to stay ahead of the competition.

If you’re interested, hear what LangChain’s creator has to say about its capabilities.

LangChain vs ChatGPT

We’ve already established that LangChain is a tool that helps you leverage technologies like ChatGPT to create apps and experiences. But let’s do a quick comparison just to clarify.

ChatGPT:

Ask it a question, get an answer. It might be brilliant, it might be bullshit, but you’ll get an answer.

LangChain:

Ask it a question, get a whole lot of nothing. That’s because LangChain is a tool to help build things—it’s not a chat app.

… … 

ChatGPT:

Gives you general-purpose conversational AI that can be used out-of-the-box for nearly anything you can think of.

LangChain:

Gives you the building blocks to create custom AI applications for nearly anything you can think of. 

… … 

ChatGPT:

You’re the end-user.

LangChain:

You’re making things for the end-user.

… …

ChatGPT:

Super easy. If you can type, you can use it.

LangChain:

Super not-so-easy. If you can type, you’re still gonna need to learn how to build apps to use it. But that’s OK, because…

Remember: the goal here isn’t necessarily to learn to make apps with LangChain. The goal is to understand what’s possible with it so you can design the user experiences of tomorrow.

Let the devs handle the code while we designers explore what’s possible with this new technology.

You can see more about what exactly LangChain is here

The UX possibilities of LangChain

Chatbots that don’t suck

A tool like LangChain can take our chatbot game from zero to chatbot hero. A LangChain app can have a much more sophisticated understanding of context compared to plain ol’ gen-AI like ChatGPT. This means more sophisticated chatbots with a deeper understanding of the nuances of a conversation. So users will have a more natural and less frustrating experience. 

Tech like LangChain will help push chatbots over that hump so they become truly useful by seemingly understanding the user at the same level as a human customer support rep. 

Lightspeed summarization and analysis

A LangChain app can summarize a massive amount of information at incredible speeds. This could revolutionize how we present information to users, allowing for dynamic content adaptation based on user preferences or time constraints.

If you’re not familiar with Slack’s AI recap feature, it creates customized recaps of use-selected channels and groups. With tech like LangChain and the right LLM, you could easily add a feature like this to give users summaries of long conversation threads, lengthy documents, or even entire chat histories.

Check out this case study on building a text summarization app with LangChain.

Hyper-personalized recommendations

Products like Spotify, Netflix, and Amazon already offer sophisticated personalization, but they use similarities between users to make recommendations. Basically they just categorize users. 

But tech like LangChain can bring it to a higher level with advanced natural language processing and data integration capabilities for hyper-personalized experiences. It means the AI will learn about you as an individual, and not just put you into a bucket with similar users, for eerily accurate suggestions.

Semantic search

LangChain can power advanced search functionalities that understand the intent behind a user's query, rather than just analyzing the keywords. 

This could dramatically improve how users find information within applications. Maybe products like Slack, Jira, Confluence, and others now offer enhanced search functionality that grasps context and intent, making it easier to locate specific messages, files, etc. within the product.

So basically, the UX search is changing before our eyes

Multi-modal Interactions

Finally, leveraging LLMs can give users a seamless experience when using both traditional inputs (clicking, typing, etc.) and more natural language-based ones. 

For example, Adobe Firefly lets users generate images by simply describing what they want in plain language. But then, users can use traditional design interfaces tools to refine the results. This powerful multi-modal approach can make users feel like Superman for an incredible UX experience.

Embrace the wave…

… Or be left behind. That’s how I see it. Staying on top of the latest technological developments help us build better digital products and experiences. 

And to be sure, LangChain is just one of many frameworks being built to leverage the power of generative-AI. There are many other new and exciting advances coming down the pipe.

So, what’s your take on leveraging AI for better UX? Any ideas for how a technology like LangChain might be used to “wow” our users with a mind blowing experience?

Let me know!

Last week we saw how deep the rabbit hole can go when we used generative AI to write actual code to create a working software application.

This week, we’ve got another practical AI application for you, but much quicker and easier. 

In we go…

I asked my podcast editor how much for a new intro: voiceover, music, that kind of thing.

“$80 for a real voiceover and music. That’s $40 for the voiceover, and another $40 for the music and license.”

“Or, you could use ElevenLabs to make an AI voiceover, so then it’d be just $40 for the music.”

Ding ding ding, someone said the magic acronym. I’ve played around plenty with audio AI tools, so I instantly set the goal of doing the whole thing with AI and saving myself 80 bucks along the way. We’re talking script writing, music composition, and voice synthesis—the whole shebang, AI style!

Here’s how I did it.

Steps 1 and 2: Crafting the script and creating the music (a simultaneous process)

I started with the two elements simultaneously to save time: the script and the background music.

Step 1: The script

For the script, I turned to ChatGPT and Claude. After some back-and-forth and refinement, I settled on this version from ChatGPT:

"Welcome to 'the AI Experience Hub,' your go-to podcast for exploring the future of AI in design. Join us as we dive into conversations with industry leaders, uncovering the latest trends and insights in building groundbreaking AI application experiences. Crafted by humans, for humans, with the power of AI—let's dive right in!"

This concise version perfectly captured the essence of the podcast.

OK, getting a few sentences from ChatGPT impresses exactly no one at this point—I get it. So how about that music? 

Step 2: Creating background music with Suno AI

For the music, I used Suno AI, a groundbreaking AI-powered music generation tool that's poised to revolutionize the music industry. Suno allows users to create original, high-quality music simply by describing the style and mood they want.

Just a year ago, AI-generated music was mostly cacophonous musical gibberish. It was basically unlistenable. But not anymore. 

Suno represents a significant leap forward in AI-generated music. Unlike traditional music production, which requires extensive knowledge of instruments, composition, and audio engineering, Suno democratizes music creation. With just a text prompt, anyone can produce professional-sounding tracks in various genres.

Suno’s potential impact on the music industry is profound:

1. Accessibility: It opens up music creation to those without formal training or expensive equipment.

2. Efficiency: Composers and producers can quickly generate ideas or backing tracks, streamlining the creative process.

3. Customization: Businesses and content creators can easily obtain rights-free, tailored music for their projects.

4. New creative possibilities: The tool can inspire unique compositions by blending genres or creating entirely new sounds.

5. Disruption of stock music: The ease of creating custom tracks could challenge the traditional stock music industry.

But while Suno and similar AI tools are powerful, they're not replacing human musicians. Instead, they're becoming new instruments in the creative toolkit, augmenting human creativity rather than supplanting it.

For my podcast intro, I prompted Suno with, "Synthwave - Retrowave upbeat podcast intro." The result was a catchy, futuristic tune that perfectly complemented the podcast's theme, created in a fraction of the time it would take to produce or license a similar track traditionally.

Here’s the result:

https://suno.com/song/096f255a-0605-4868-a774-94b4b93fe556

Step 3: Voicing the intro

For the voiceover, I used ElevenLabs, a cutting-edge AI voice synthesis platform. ElevenLabs uses advanced machine learning to generate highly realistic and customizable voice content. After trying a few voices, I went with "Bill." And though Bill’s voice still has a slightly robotic quality (especially in the speech cadence), I thought it fitting for an AI-focused podcast.

Here’s the result:

https://jmp.sh/unNv7vNt

Step 4: Combining the audio

The final step was to combine the background music and voiceover. 

For this, I used Kapwing, a free online video editor that also handles audio mixing. It was pretty straightforward. I pretty much just dropped in the two audio clips, slid them left or right a bit for the timing, and then adjusted each one’s volume to get the right blend.

The result

The entire process—from writing the script to generating music, voicing the intro, and combining the audio—took under 3 minutes. Seriously.

And the end product is a professional-sounding podcast intro that captures the essence of "The AI Experience Hub."

Here it is:

https://jmp.sh/uJ6RnsgW

Recap

So yeah, it only took me a few minutes and saved me $80, FTW. I also like the scrappiness of it, especially for a podcast all about the power of AI. Seems appropriate.

I’m chalking this one up as another AI win. In the time it takes to microwave a burrito, I whipped up a pro-level podcast intro that would've typically involved several people, lots of back-and-forth emails, and real hard-earned cash. Is this the AI productivity boost Wall Street has been talking about? 

And once again we see, it’s not the AI overlords pulling the strings. All those prompts, tweaks, and "Nah, that's not quite right" moments? That was still a flesh-and-blood human! So we're not just backseat button-pushers—we're the conductors of this AI orchestra.

Claude's first attempt was about as on-point as a drunk karaoke singer, but that's the beauty of this whole process. We get to cherry-pick the good stuff and then polish it until it shines.

Thank you for reading.

I asked ChatGPT to write lyrics and play a song only for you:

(Verse)

Thanks for the read, you’re the best,

Podcast intros, you’ve aced the test.

With AI’s groove, we’ve set the vibe,

Podcast intros come alive.

(Chorus)

Big thanks, big shout,

For hanging out.

Podcast intros, you rock,

You make us whole.

Plot twist: We're launching a podcast!

Surprise! If you haven't put two and two together yet, all this AI podcast intro wasn't just for kicks. We're launching "The AI Experience Hub" podcast! It's your front-row ticket to the AI revolution in design, packed with chats with industry bigwigs and insider tips on crafting mind-blowing AI experiences.

Stick around for our debut episode, where we'll spill all the tea on how we cooked up this podcast with our AI sous-chefs. Trust me, it's gonna be wilder than a rollercoaster ride through the metaverse.

So, what's your take on this AI content creation madness? Have you dipped your toes in these digital waters? Spill your guts in the comments below! And hey, while you're at it, hit that subscribe button for "The AI Experience Hub." Your ears will thank you for all the AI and design goodness coming their way!

Focus on the process, not the tool

11:30 AM, BREAKING NEWS: Figma announces Figma AI—the crowd goes wild! 

11:35 AM, UPDATE: Figma withdrawals feature due to a potential lawsuit over designs that looked like the Apple weather app.

That was quick!

And to be honest, Figma AI has been around for more than a year now, and it’s mostly useless, generating mind maps that don't make a lot of sense. Figma’s AI announcement was mainly a marketing thing.

For me, the lesson here is to never fall in love with a tool (looking at you Invision), but rather, focus on a process. So that’s what I did when I decided to tackle the project outlined below.

No-code + AI = code!

I’ve been building no-code apps for a few years now. But recently, I thought about how AI is set to change the app design process, and I decided to branch out a bit—from pure no-code to no-code + AI-generated code.

Yes, it is possible to create viable, working apps using AI-generated code. Here’s what I did and learned how to use Claude’s new Artifacts feature along the way.

Creating an AI dermatology app

One day, while pondering the intersection of healthcare and technology, an idea struck me: what if we could leverage AI to help people assess potential skin issues from the comfort of their homes? What if I could build an MVP for this?

I set out to build DermaLog, a skin analysis app that would use OCR (Optical Character Recognition) and AI to analyze images of skin conditions. So users simply upload a picture of a skin irregularity and get back a likely condition.

And remember—I don’t code, at all. Sounds like a daunting task? That’s what I thought too… at first.

Here are the steps I took.

Step 1: initial prompts

I started with Claude, an AI language model developed by Anthropic. I began with a simple prompt, typos and all:

"I want to create a python app that is based on OCR library that could take an image of skin and then use the api of open ai in order to analyze if there is any skin issue in that picture and medically analyze it. Then I need the option to prompt the data to a doctor if there is anything suspicious or not so I need it to tag it as dangerous or not dangerous. For now, tell me the architecture of it."

Claude's response was comprehensive, outlining the main components of the app, suggesting libraries to use, and even proposing a database structure. This initial interaction set the stage for our collaborative development process.

I followed this with a second prompt: “Let’s do it!”

Step 2: setting up the Flask application (what the heck is Flask?)

The development process unfolded through a series of prompts and responses. At one point, Claude casually mentioned that I should use Flask. 

OK, that’s cool. But, "What is Flask?" I asked. 

Claude provided a concise explanation (it’s a web framework). With this new knowledge, I prompted Claude to generate the necessary code for the basic Flask application structure. I then copy/pasted the code for these files:

run.py

init.py

routes.py

As we progressed, my prompts became more specific:

"Change the name to DermaLog. Make sure to create these colored tags based on how dangerous the findings are. Make sure to add % of what it might and how accurate the results are to be and add different suggestions to it if possible."

Claude responded with updated HTML and Python code, incorporating these new features and design elements.

By the end of our initial development session, what started as a vague idea had evolved into a functional prototype for DermaLog.

I had a clear understanding of the app's structure, its key features, and even some insights into potential future improvements.

As I prepared to refine and test the app further, I couldn't help but marvel at how AI had enabled me to bring my idea to life without writing code myself.

I had crossed the threshold and stepped into the world of app development democratization. 

Step 3: image upload and AI analysis

I asked Claude how to create a form for users to upload images. It spit out both HTML and Flask route code, which I copied directly into my project.

OCR Processing: When Claude suggested using Pytesseract for OCR, I said, “Claude, if I didn’t know what Flask was, do you really think I know what Pytesseract is? I can’t even pronounce that!”

OK, I didn’t really say that—I just asked for clarification on what OCR was and how to implement it. Again, I copied and pasted the code Claude generated.

For the image analysis, we initially attempted to integrate the OpenAI API, but encountered persistent issues. When I shared these problems with Claude, it suggested using its own API instead. Jealous much, Claude?

This led to a significant pivot in the approach, and I had to ask Claude to generate entirely new code for this integration, which I again pasted into the app's code.

For result classification, the system was also entirely generated by Claude based on my description of what I wanted (categories like: Safe, Needs monitoring, Potentially dangerous).

Step 4: enhancing the UI

For the UI, I described to Claude what I wanted, and it generated the HTML and CSS for the homepage and the results page.

Step 5: overcoming challenges

Developers know that debugging is a huge part of the job. Good thing Claude can debug too! 

E.g., when I shared a "TemplateNotFound" error, it guided me through debugging steps, correcting the code.

Same thing for persistent problems with API authentication. It took a few passes, but Claude fixed those too. It even cleaned up the Python indentation errors, which it turns out are a thing. (Did I mention that I don’t code?)

Step 6: adding advanced features

Even as we added more advanced features, our partnership of director and coder remained intact. 

For example, I asked Claude how to add a WhatsApp link, and it provided the HTML code. I added Percentage Likelihood and Accuracy Assessment features too, which again were created by prompt-generated code.

At this point, it should be clear that my role was nothing at all like that of a traditional developer. Instead, I was more of a project manager, outlining requirements, asking questions, and making decisions about the app's direction. 

Claude served as both teacher and coder, explaining concepts, generating code, and solving problems. 

Step 7: testing

I asked Claude, "How do I run the Flask app and test it?" Claude provided me with the command:

python run.py

I copied this into my terminal, and to my surprise and delight, the app started running on my local machine. However, this was just the beginning of our testing process.

That’s how the copy/paste terminal looked like, everyone have this one their mac.

Step 8: more bugs

As I started interacting with the app, I encountered several issues: image upload problems, OCR functionality, and API issues. Claude handled them like a boss, e.g. suggesting I update the API key and modify the request headers. Each suggestion came with code snippets. Copy, paste, fixed!

Step 9: more UI improvements

As the core functionality stabilized, I turned my attention to enhancing the user experience:

1. Error handling: I asked Claude how we could provide better feedback to users when something goes wrong. It suggested implementing Flask's flash messages and provided code to display these messages in our HTML templates.

2. Loading indicators: To improve perceived performance, I asked Claude how to add a loading spinner while the image is being analyzed. As expended, it provided HTML, CSS, and JavaScript to implement the feature.

3. Responsive design: I noticed the app didn't look great on mobile devices. Claude gave me additional CSS to make the design responsive, which I added to our stylesheet.

Step 10: refining the AI analysis

To improve the accuracy and usefulness of DermaLog's analysis, we made several iterations:

1. Adjusting prompts: I experimented with different prompts for the Claude API, asking the AI assistant to help me craft more effective instructions for skin analysis. We went through several iterations, fine-tuning the language to get more detailed and accurate responses.

2. Implementing confidence levels: I asked Claude how we could include confidence levels in our analysis. It suggested modifying our API call and results parsing to include this information, providing the necessary code changes.

3. Adding disclaimers: On Claude's suggestion, we added a disclaimer to remind users that the app's analysis is not a substitute for professional medical advice. Claude provided the text for this disclaimer and the code to display it prominently in the app.

Step 11: performance optimization

As a final step, I asked Claude about ways to optimize the app's performance. It suggested several improvements:

1. Image compression: Claude provided code to compress uploaded images before analysis, reducing API call times.

2. Caching: We implemented basic caching for API responses to improve performance for repeated analyses of the same image.

3. Asynchronous processing: Claude suggested and provided code for implementing asynchronous processing of image uploads and API calls, improving the app's responsiveness.

This iterative process of testing, identifying issues, and implementing solutions allowed me to significantly refine and improve DermaLog.

Meet DermaLog

Guess you probably want to see it by now. 

Check out the demo video

Believe me, I’m not trying to say this app is perfect and ready to be used as-is. But it does show what’s possible with AI-generated code, even if you have no clue about coding. 

Turns out that a heck of a lot is possible!

DermaLog serves as a proof of concept. It's perhaps enough to get buy-in from clinics and attract investment to develop a more comprehensive, secure, and accurate product.

And it was all done in the span of a few hours while holding a one-month-old baby in my lap. 

It’s your turn

So, who among you is brave (or crazy) enough to embark on a similar journey? What are your ideas for apps that could be rapidly developed using AI-generated code? 

Send me a link to anything you create in a similar process (and how you did it) and I’ll feature it in an upcoming edition.

Happy coding… err, I mean, happy prompting!

Marc Andreessen coined the phrase "software is eating the world" in a 2011 WSJ essay during a time of rapid growth in the tech industry, when software companies were gaining significant valuations.

Andreessen, a prominent venture capitalist who co-founded Netscape, argued that despide the tech bubble burst of the late ‘90s, major tech software companies were now poised to disrupt and transform traditional industries, leading to a significant shift in the global economy.

He was, in a word, correct.

Today, software is everywhere—it delivers food and goods, manages transportation, healthcare, business, media, and basically everything.

And all those who failed to envision the success of Facebook, Google, Twitter, etc. and other tech giants missed out big time.

Yet little more than a decade after Andreessen's essay, another technology has emerged with a disruptive potential that’s perhaps several scales of magnitude larger—it’s AI’s turn to eat the world.

But will the nom-nom-nom of generative AI turn out to be just a nibble, or will it be the colossal chomp investors are betting on?

Let’s “dig in” and see if we can’t start to make some sense of it all.

The biggest AI winner so far

During a gold rush, sell shovels. That’s precisely what Nvidia, the top manufacturer of shovels GPUs, is doing—selling companies the technology that makes AI applications possible. 

Image prompt for Midjoruney: vintage group portrait, gold rush prospectors from various backgrounds, lined up with their tools, authentic 19th century mining camp backdrop, high detail black and white, historical accuracy, 1860s wet plate collodion style

And it’s going well: at its peak on June 18, 2024, Nvidia was briefly the most valuable company in the world with a market cap of $3.34 trillion (with a T), surpassing Microsoft, Apple, and Google.

But will Nvidia soon meet its untimely end, going down in a spectacular blaze of glory à la Pets.com?

Unlikely—though a price correction is possible. 

With Nvidia leading the charge, AI hype has been bolstering US stocks since early 2023, helping lift the global economy out of its post-COVID blues, as investors anticipate that massive productivity boosts will turn into increased revenue.

• Market cap history of NVIDIA from 2001 to 2024

AI will assist, not replace

Black Eyed Peas frontman will.i.am sees AI as a tool to enhance human creativity, rather than replace it. I couldn't agree more. 

For example, though AI helped me to create this post (Perplexity to quote resources, Chatgpt and Claude to wrap my head around concepts) it was still edited and partially co-authored by an actual human (shout out to the talented Aaron Raizen who I’m pretty sure isn’t a robot).

Today’s generative AI can recycle ideas—bashing existing data points into one other—but it can’t really craft coherent new ideas, for example the concept that AI is not necessarily eating the world.

At least, it can’t do it yet.

Nearly everyone who’s added AI to their day-to-day work tasks says that it changed how they do some things, boosted productivity, improved processes, maybe even added more work.

But it hasn’t replaced very many people's day-to-day work. It may replace full customer support teams one day, but, for example, it also helped the company Klarna increase revenue and efficiency by exactly… a whole bunch!

• Case Study: Klarna’s Revolutionary Use of AI in Customer Service and Operations

• Klarna sees its AI assistant as a success. Is it too early to tell?

Is it going to be like this forever? Doubtful, but for now, the human factor is still critical and most of the “AI will steal our jobs” fear comes from people with outdated worldviews.

You know, like back when you could be inefficient, slow, but still make a ton of money. (“Those were the days my friend, we thought they’d never end.”)

If we look at law firms, for example, we see their work has barely changed for decades. And though you didn’t hear it from me, some say the top firms that made the big bucks did so through a network of long standing connections—and by having younger interns busting their humps for peanuts.

That system was and still is broken.

So if gen-AI continues to advance to the point where it replaces most lawyers, would I be upset? Heck to the no.

Their work probably wasn’t that important if it was so repeatable. If they become a horse pulling a cart in a world full of Ford Model Ts, then it’s a good thing. 

Image prompt for Midjoruney:: vintage advertisement style, Model T Ford centerpiece next to a horse, white background, 1900s graphic design elements, bold typography 'The Car for the Great Multitude', high detail on vehicle, muted color palette

I mean, can you imagine a world without lawyers? Sounds OK to me. 😂

Consulting will never be the same

Consulting firms, including the Big Four, are rapidly expanding into industries like law, leveraging AI for growth and transformation.

AI has become essential, enhancing efficiency, helping develop new solutions, and improving client services.

The future of consulting, and perhaps many other fields, will be one of collaborative intelligence, combining human creativity with AI scalability.

These firms have easy access to enterprise clients and are looking to make—and are already making—billions from gen-AI by partnering with leading AI startups for industry-changing solutions.

For example, PwC's partnership with Harvey aims to introduce an advanced legal AI solution, and there are probably 2000 similar solutions being developed around the world right now.

Legal firms will have difficulty resisting AI's impact, raising the question: should they even try, or should they just embrace it?

• PwC announces strategic alliance with Harvey, positioning PwC’s Legal Business Solutions at the forefront of legal generative AI

• How AI is Reshaping the Consulting Industry

What if AI is actually eating the world?

When software ate the world, it gave us services like Uber that “disrupted” transportation and food delivery, or Airbnb and Booking.com that transformed lodging and accommodations.

We got things like better medical documentation, streamlined financial transactions, personalized entertainment recommendations, and so much more. 

Overall, the wealth of humanity, as well as hunger and healthcare issues, have all improved.

Software ate the world, and well, it was kinda great. Turns out the world can be pretty tasty.

So even if AI is eating the world, like software before it, maybe that isn’t necessarily a bad thing.

Maybe AI’s chomp-chomp will be a win-win for us all.

Maybe it will help us explore the galaxy, find a cure for cancer, and create a completely new understanding of who we are and what we do here.

Bon appetit!

As a game enthusiast and ux person with more ideas than coding skills, I recently stumbled upon a new opportunity: creating a Snake game with the help of the AI assistant of Claude (a popular LLM).

Curious about the potential of AI in game development, I decided to put it to the test. What followed was a cool adventure of turning a simple concept into a playable game in just minutes.

Here's my hands-on journey of collaborating with Claude to bring Snake to life, from start to finish.

Step 1: Initial Code Generation

I started by prompting Claude: "Create a basic Snake game using JavaScript and HTML." Within seconds, I received a complete code snippet.

I created a new HTML file, pasted the code, and opened it in a browser. Surprisingly, I had a functional, albeit basic, Snake game.

It had a scoring system, very basic graphics, and movement mechanism, all created in seconds.

Step 2: Understanding the Code

Despite my limited coding knowledge, I could grasp the general structure:

• HTML canvas for rendering

• JavaScript for game logic

• Functions for moving the snake, generating food, and handling collisions

Checking out the code itself allowed me to reverse engineer what was done so I could read through the code and understand exactly what it did, and also make changes if needed.

In addition to that, the AI assistant also kindly explains how exactly it works.

This immediate output allowed me to focus on design aspects rather than getting stuck on syntax.

Step 3: Iterative Design

With a working prototype, I began iterating on the design:

1. Color Scheme: I prompted, "Change the snake color to neon green and the background to dark gray." The AI quickly provided the necessary code changes.

2. Food Appearance: I asked, "Make the food look like an apple." The AI suggested using a simple red circle and provided the code to implement it.

3. Minor changes: Creating new graphics including a snake-like head image using AI. Surprisingly, it easily incorporated these assets. To increase difficulty, I also added apples of various sizes spread across the UI.

Step 4: Enhancing Gameplay

To improve user experience, I made several gameplay adjustments:

1. Speed Increase: "Make the snake move faster as the score increases," I suggested. The AI provided code to gradually decrease the game loop interval.

2. Sound Effects: "Add a sound effect when the snake eats food," I requested. The AI included audio functionality with a sample sound file.

3. Responsive Design: "Make the game responsive to different screen sizes," I asked. The AI adjusted the canvas sizing code to be more flexible.

Step 5: Testing and Refinement

I played the game repeatedly, noting areas for improvement:

1. Controls: The original arrow key controls felt clunky. I asked the AI to "Implement smoother controls with key press buffering." This resulted in more responsive gameplay.

2. Visual Feedback: "Add a brief flash effect when the snake eats food," I suggested. The AI provided a simple animation using canvas drawings.

3. Difficulty Levels: "Create easy, medium, and hard difficulty options," I requested. The AI implemented a difficulty selector that adjusted starting speed and score multipliers.

Step 6: Polishing the User Interface

For the finishing touches, I focused on the UI:

1. Start Screen: "Design a start screen with game instructions," I prompted. The AI created a simple overlay with game rules and a start button.

2. Typography: "Use a retro-style font for the game title and score," I suggested. The AI provided CSS to import and apply a pixel font.

3. Animations: "Add subtle animations to the start and game over screens," I requested. The AI implemented fade-in effects using CSS transitions.

Reflections on the Process

This hands-on experience with AI-assisted game design was enlightening:

1. Rapid Prototyping: I went from concept to playable game in minutes, not weeks, not hours.

2. Focus on Design: Offloading coding to AI allowed me to concentrate on user experience and game feel.

3. Learning Opportunity: Through iterative prompts, I gained insights into game development principles.

4. Limitations: While powerful, the AI sometimes provided solutions that needed tweaking for optimal performance.

5. Creativity Amplified: The quick implementation of ideas encouraged more creative experimentation.

AI HTML Snake game version 1:

AI HTML Snake game version 10:

Conclusion

AI-assisted development dramatically accelerated the game design process, allowing me to create a fully functional Snake game despite my limited coding skills. This approach enabled rapid iteration on design ideas and gameplay mechanics, resulting in a polished final product.

For UX designers looking to prototype interactive experiences quickly, AI-assisted coding tools offer a powerful way to bridge the gap between imagination and implementation. While not a replacement for deep coding knowledge, it's an invaluable tool for turning design concepts into reality efficiently.

You can play my new snake game here: https://yuvkesh.github.io/snake/

Large Language Models (LLM), like ChatGPT and its brethren, are powerful tools. This we know. 

Yet for all their usefulness, they still have a major problem—they’re liars. Damn good ones too.

Ask them nearly anything, and they’ll likely spit out a well-constructed, coherent, and conversational answer. But that answer has a high probability of being partial or total BS. 

That’s because LLMs like ChapGPT are designed to be convincing, but not necessarily factual. Part of this stems from the fact that they’re trained on closed datasets, with ChatGPT’s knowledge ending in 2022.

But there are technologies addressing this problem, and as tech workers, it behooves us to understand them.

You already know LLM and GPT, so make room in your brain for yet another AI-related, three-letter acronym.

Allow me to introduce you to RAG.

What is RAG?

RAG stands for Retrieval Augmented Generation. It’s a framework/process/technique that makes LLM responses more reliable and accurate by retrieving related and up-to-date data that’s relevant to a user’s query.

It combines the power of both retrieval- and generative-based AI models to produce better results.

Think of it this way.

Imagine you’re in a library, looking for a specific book. Not only do you have access to the stacks, but there’s another helpful resource that makes the process easier—the librarian.

That librarian can not only guide you to the right section, but also help you find related material, or even give you a summary of resources tailored to your interests. This is a bit like how RAG makes an LLM more effective.

Or, think of it like a plug-in.

Figma on its own is an excellent piece of software, but layering on your favorite plug-in makes it even better.

RAG is like a plug-in for LLMs—an additional layer that enhances their capabilities and usefulness.

Components of a RAG system

The RAG technique or process has three major components: retrieval, generation, and data sources. Here’s a quick breakdown.

1. Retrieval

• Makes use of vector databases, which store data in a way that allows for efficient retrieval based on the similarity of information.

• Semantic search: utilizes natural language understanding to fetch the most relevant documents or pieces of information from the database.

2. Generation

• Large Language Models generate human-like text based on the retrieved information, ensuring that the output is coherent and contextually appropriate.

3. Data sources

• Documents: includes textual data such as articles, research papers, and books.

• Databases: structured data sources like SQL databases.

• APIs: real-time data from various web services and applications.

Typical workflow of a RAG system

1. User query: the process starts when a user inputs a query or request.

2. Retrieval: semantic search is used to find the most relevant information from the vector database and other data sources.

3. Augmentation: the retrieved information is then fed into the generation component (the LLM).

4. Generation: the LLM processes the information and generates a coherent response tailored to the user's query.

5. Output: the system delivers the generated content to the user, providing a relevant, personalized, and (hopefully) accurate response.

Diagramming the integration of retriever and generator components from “A Simple Guide To Retrieval Augmented Generation Language Models”

Rag in UX

So how can RAG help us create better user experiences?

Imagine you walk into your local coffee shop. The barista sees you and not only remembers your usual order, but also suggests a new blend you might enjoy based on your past orders and even your current mood. 

RAG can help us deliver this kind of personalized experience efficiently and at scale—much better than an LLM like ChatGPT can on its own.

So let’s explore some possibilities for RAG in UX across different fields with examples and use cases.

UX use cases for RAG

Customer support chatbots

Example: A chatbot integrated with a company's internal knowledge base to provide accurate and up-to-date responses to customer queries.

Use case: Enhancing customer service by retrieving relevant information from knowledge base texts, policy documents, FAQs, user manuals, etc. to give users specific and accurate answers to their questions with incredible speed and accuracy. 

E-commerce

Example: An internal system that can access and analyze product data, user reviews, and individual user data.

Use case #1: Better, faster, and more accurate product descriptions across the site.

Use case #2: User-specific, customized product descriptions based on purchase and browsing history. The system could change the order of images and product details at the level of the individual user, showing users the things that matter most to them first. 

Medical assistants

Example: A generative AI model supplemented with a medical index.

Use case: Assisting medical staff by providing precise information from medical databases, research papers, and patient records to support clinical decisions. 

Deploying RAG systems in this context could have far-reaching benefits, reducing medical error and thus reducing medical malpractice suits and lowering insurance costs.

Financial analysis tools

Example: An AI assistant with access to market data and financial reports.

Use case: Helping financial analysts make more informed decisions by retrieving insights from up-to-date market trends, financial statements, and economic indicators, and then generating actionable insights.

Employee training and productivity

Example: AI systems that access technical manuals, training videos, and logs.

Use case: Enhancing employee training programs and improving productivity by providing contextual information and step-by-step guidance based on the latest company protocols and procedures.

Legal research assistants

Example: A legal AI assistant that retrieves case law, statutes, and legal precedents.

Use case: Supporting lawyers and judges by providing relevant legal information and precedents to aid in case preparation and decision-making.

Vacation rental information systems

Example: A chatbot providing contextual information about vacation rentals.

Use case: Offering detailed, fact-based answers about amenities, accessibility, and local attractions for potential renters.

RAG for a UX content style guides

The above are some hypothetical or potential use cases for RAG in UX. But here’s a content style guide RAG app (slow loading) we’re already implementing for one of the client of the UX Writing Hub (the UX training and agency company I’ve been running for the past 6 years).

This RAG app has been trained on a style guide for a client named Beti, a cloud-based platform for construction site management.

It allows the people at the team of Beti to easily query the style guide to get answers on how to follow it.

On the front end, it seems like good old ChatGPT, while on the back end, it's a trained RAG that can answer questions related only to the style guide itself, and it gets almost zero hallucinations and knows how to generate to a piece of microscopy based on the style guide rules.

Next week, I’ll share a step-by-step case study for building this type of RAG.

RAG in UX summary

We stand at the cusp of a technological upheaval whose scale and impact we can’t yet fully perceive.

LLMs like ChatGPT changed our world seemingly overnight. But when coupled with other technologies, the possibilities will only continue to expand.

As UX professionals and technology workers, it’s our duty to not only keep up with the latest developments, but to constantly consider how their implementation can affect the user experience. 

So, considering what you now know about RAG, how might you implement this type of system into the products you work on? 

Let me know! 

Refer a friend

Written by Yuval Keshtcher, Edited by Aaron Raizen

Welcome to the AI Experience Hub: Your Gateway to Winning AI Experiences

Greetings, fellow explorers of the digital realm! Welcome to the AI Experience Hub, where your journey to creating winning AI experiences begins. Whether you come from a background in UX and product design or you are simply curious about this new technological dimension that's sweeping the world, you are in the right place.

This is not just about learning the mechanics of AI but about understanding how to weave AI into experiences that delight, engage, and innovate. It's about crafting journeys where every interaction feels intuitive and every solution feels smart.

At the AI Experience Hub, we'll teach you hands-on technologies using tools such as Make, various large language models (LLMs), and more.

Our goal is to help you build user-centric AI applications that demonstrate the real value of AI and UX. By integrating these cutting-edge tools into your projects, you'll gain practical insights and skills to create impactful digital experiences.

Our mission at the AI Experience Hub is to equip you with the knowledge and skills to design AI experiences that not only meet user needs but exceed their expectations. We believe that a strong foundation in UX principles, combined with a curiosity for the cutting-edge capabilities of AI, can lead to truly transformative products.

The AI Experience Hub is created by Yuval Keshtcher, the founder of the UX Writing Hub, one of the biggest UX writing training and agency companies in the world. The UX Writing Hub works with the biggest brands and tech companies globally, shaping user experiences for millions. With Yuval's expertise and vision, the AI Experience Hub aims to be a beacon for those looking to harness the power of AI in their work.

So, take that first step. Dive deep into the world of AI with us. Together, we’ll explore the latest trends, tools, and techniques that are shaping the future of user experiences. Let’s create something extraordinary.

Welcome to the future. Welcome to the AI Experience Hub.

Subscribe now