Hi folks!👋🏻 This is The Prompt! So many things to share today!

Let's get it

FEATURED

AI Wearable that can see the world 🕶

Project Ring is a wearable AI, that’s multi-modal. It can take photos, listen to voice commands, give back answers or just chat with you about what “it” sees.

It can:

• 👂 hear using OpenAI’s Whisper,

• 👁️ see using an image-to-text model, and

• 👄 speak using text-to-voice using ElevenLabs models.

How is it built?

It has three components.

• A ring: Can take photos, & trigger voice commands connected to a

  Raspberry Pi Zero.

• A phone app: It records the voice commands, and plays back the answer from the AI.

• A cloud app: That handles AI processing in the cloud for all 3 models that are used

The punchline?

It was all coded by GPT-4. All 750 lines of code.

The author says that he has a developer background, but that he hasn’t coded for years.

“I can now say that it is possible, but not easy, to create software prototypes entirely using GPT-4”

It’s definitely interesting to see how GPT-4 can augment coding, and how human-machine interaction could be possible outside of our browsers.

Watch the demo here, or read how he built it.

NEW TECH

• Ask questions to your docs without an internet connection with PrivateGPT. It’s 100% private, and no data leaves your environment. You can upload documents and work with them.

• Open-sourced Code Interpreter. If you don’t have access to the Code Interpreter in ChatGPT, you can use this open-source implementation.

• Drag your GAN: Image editing model. 

  With this model, you can deform an image with precise control if you move the pixels, and manipulate the pose, shape, expression, and layout of the generated image.
  
  I guess - Photoshop is so 2000 and late?
  
  No demo is available, but they’ll publish one soon. It works really well! See for yourself 👇🏻

WHAT ELSE IS GOING ON

🦙 Apple pre-announced Personal voice. You just read 15 minutes of phrases and it will generate an on-device AI model of your voice. Designed for users at risk of losing the ability to speak, but it will be available for everyone (probably in Sept). Who said that Apple was falling behind?

👀 Official ChatGPT iOS app is here. Better UI for sure, but apart from just chatting with ChatGPT, you can use your mic to talk to it as well.  

🤖 AI robots are coming. Sanctuary AI revealed Phoenix, which is their humanoid robot designed for the workspace. And hours after their launch, Tesla shared a tweet about their humanoid robot Optimus.

RESOURCES

The best resources we came across lately that will help you become better at writing prompts & building AI apps.

📚 Vector embeddings - simple explanation [short very informative YT video]

👋🏻 How to use and build with Lanchain [ a must-watch]

🎥 How to use the remix feature in Midjourney [Twitter thread]

TOOLBOX

The latest AI tools to use or get inspiration from.

• Poetry tips: Generate poems from descriptions

• NameWizard: Domain names with the help of AI

• Yoodli: AI speech coach

• Langdock: Convert any API into an LLM plugin

• Seodity: Content generator with SEO data

• Brandsnap: Discover domains & social handles

PROMPT OF THE DAY

TOOL

Midjourney

PROMPT

1960s French New Wave Cinema, experimental film still, side-angle view of an angsty male protagonist browsing produce in post-World War II society, intimately locking eyes with a tomato, blue night on the Paris streets, high contrast dramatic shadows::

RESULT