👏 ChatGPT Voice hack & create web apps

AI-HOI AInauts,

Welcome to the latest issue of your favorite newsletter. Today with helpful practical tips to follow and the most important news. Here's what we have in store for you:

• 📱 The ultimate ChatGPT voice hack: Turning your ideas into structured text

• 🚀 Walkthrough: Create your own web app and bring it online

• 📰 AI news quickie: The HAIlights from the industry

Here we go!

📱 The ultimate ChatGPT Voice hack: Turn your ideas into structured text

Does this sound familiar: You have a flash of inspiration but can’t write it down properly? Or are you bubbling with ideas, but your fingers can't keep up with typing? If this is the case, we have a really cool hack for you today.

Speak it, and have ChatGPT take notes!

Imagine if you could just babble away and end up with a perfectly worded text, a crisp summary and even a to-do list - all without lifting a finger!

How it works: Your step-by-step guide

1. First install the ChatGPT app on your phone or desktop. Select the desired voice and activate the "Background calls" feature on your phone.

2. Start a new chat with this prompt (or use our GPT *just tell it to act in your desired language):

You are my helpful assistant who transforms my spoken thoughts into a well-written text. 

I start talking and you should ONLY say the word 'Continue' if I pause too long (nothing else). As soon as I say 'Done', 'Stop' or 'End', I want you to turn my thoughts into a well-written and clearly structured text. 

Then please summarize the main points and extract all the tasks from our conversation.

IMPORTANT: As long as I have not said "Done", "Stop" or "End", always answer with "Continue", nothing else, no further details etc.

3. Click on the headphone icon in the app and just start talking!

4. At the end, say "Done!", "Stop" or "End" loudly and clearly, and you will be presented with your summary.

Our take: Speech is the new typing

This approach is not only cool, it also shows us where the journey is heading: towards the seamless integration of AI into our creative process. The barrier between thinking and writing is getting thinner and thinner.

Imagine how you could use this hack in your everyday life: Lightning-fast meeting notes, spontaneous blog post ideas, drafts for documents or even the outline for your next brilliant business idea - all just by speaking! You save time and can let your ideas run free.

P.S.: You can also build your own GPT (only for Pro users), or use ours (also for free accounts). Voice is becoming more and more important, and soon we'll really have an ingenious assistant in our pocket (see the examples of the new ChatGPT Voice feature below).

🚀 Walkthrough: Create your own web app and bring it online

Do you have 15 minutes and are in the mood for a fun experiment? Then let us program an app together and publish it online.

We'll do it step by step - without writing a single line of code ourselves. Let's go!

1. Choose your coding wizard 🪄

Before you get started, you need the right tool. Claude 3.5 Sonnet is our favorite, but alternatively you can also use ChatGPT or Llama 3.1. All are available to you free of charge and can code like champions.

2. Speak plainly: formulate your vision 🗣️

Explain to your chatbot in simple terms what kind of app or website you have in mind. Be as specific as possible. Example:

I want a to-do list app that saves tasks in the browser. Use only HTML, vanilla JavaScript and a CSS library. The data must be stored in the browser, there is no server-side code.

Possible ideas would also be a flashcard learning app with digital cards, or a diary app for daily entries with mood tracking or a notebook for saving and organizing notes with a search function.

Bonus tip: Use the Claude Prompt Generator for a more sophisticated prompt. Keep in mind that your app does not have direct access to AI features (or you would need to have this built in by the user entering their API key).

3. Let the AI work its magic: code generation 🎩✨

Pop the prompt into the chatbox and be amazed at how your digital assistant conjures up a complete website for you. HTML, CSS, JavaScript - all included, without you (we) having to know a thing about it!

4. Test and expand: Local check and new features 💅

Test the output directly in Claude, then download it to test it in the browser (especially if app data is stored in the browser).

Like what you see, but want more? No problem! Go back to the chatbot and add more functions to your app.

Please note: If a single file/app becomes too extensive, Claude may lose the thread and stop in the middle. Here you can have it detach more extensive functions to separate files.

5. Publish it online: Hosting with Cloudflare 🌐

The easiest way is to simply share the code as a Claude app.

But we want to introduce you to an alternative, more independent way.

• Create a free account at Cloudflare.com.

• Go to "Workers & Pages" and then to the "Pages" tab.

• Click on "Upload Assets" and give your baby a name.

• Pack your HTML file (renamed to "index.html") into a folder on your computer.

• Drag & drop this folder into the Cloudflare interface.

• Click on "Deploy Site" and - BOOM! - the app or website is online!

• And if you want to update the app, simply click on "Create new deployment".

And you've done it! Your creation is now accessible to the whole world. Here is our example, a simple task manager app for the browser: https://taskmasterpro.pages.dev

Our take: Trial and error makes perfect!

With AI as a sidekick, you can suddenly set up your own online projects for your use cases in record time, without having to code at all. This opens up new possibilities for creativity and innovation.

Send us your creations, we are super excited to see what you come up with!

📰 AI news quickie: The HAIlights from the industry

A lot has happened, here are the most important updates from the AIniverse!

OpenAI

• OpenAI's Advanced Voice Mode is now available for the first alpha testers. X is full of examples of what it can do - here are a few of the highlights.

• OpenAI & Co. support new proposals in the US Senate that would shape the future direction of the government.

• Among other things, the U.S. Artificial Intelligence Safety Institute (AISI) is to set the direction and work together with industry.

• The fact that OpenAI is also announcing that it is granting the AISI access to new models cannot be a coincidence ...

• Microsoft surprisingly calls its $13 billion partner OpenAI a competitor in AI and search. Friendly competition or a clever move against antitrust watchdogs?

Google

• Google presents new Gemma 2 models, with a focus on security and responsible use. Incidentally, the White House is in favor of open source AI and currently sees no need for restrictions.

• Google Chrome has also received new AI features: With Google Lens, you can search for images directly in the browser, compare products from multiple tabs and find forgotten websites by voice.

• Google has sped up its Gemini chatbot and making it smarter: with the update, you can now upload files and get faster, more precise answers in over 40 languages.

• There is also an update available in the Google AI Studio and via API: the "Experimental Version (0801)" of Gemini 1.5 Pro is available - currently the #1 in the LMSYS arena for text and multimodality!

Picture & Video

• Runway goes one better: With Gen-3 Alpha, you can now create breathtaking videos from still images.

• Meta's new SAM 2 model revolutionizes AI-powered object recognition in real time - you can now segment objects in images and videos at lightning speed!

• Also from Meta/Insta comes the new "Create Your Own Custom AI" builder via AI Studio - this should take some market share away from character.ai ...

• Midjourney V6.1 has been released: Better image quality, more precise details (especially for humans), faster generation. Nice!

• Flux.1 is a new AI image generation model from Ex-Stable Diffusion developers and even promises superiority over Midjourney and Co. It’s really good, see for yourself and test it here!

• Vimeo introduces AI-powered video translation. This allows companies to offer their videos in all possible languages!

• You can now create lip-synced videos with Rendernet's Narrator - really powerful, and available in multiple languages!

• Or create your AI-Twin with the new tool from Captions.

• The Digital Hollywood AI Summer Summit Recordings are live.

Robotics & Gadgets

• The fast food chain Taco Bell is now using AI in its drive-thrus!

• The first fully automated dental procedure was performed by a robot... and your toothbrush will soon be equipped with AI too.

• Dogs have a language, ergo ... we should soon have a translator to talk to our four-legged friends.

• Figure has announced version 2 of its robot - "the most advanced humanoid on the planet".

That's it for today. We hope you liked it - see you next time!

Reto & Fabian from the AInauts

P.S.: Follow us on social media - that motivates us to keep going 😁!
Twitter, LinkedIn, Facebook, Insta, YouTube, TikTok

Your feedback is essential for us. We read EVERY comment and feedback, just respond to this email. Tell us what was (not) good and what is interesting for YOU.