😸 Meta's biggest AI model

After months of rumors, OpenAI finally pulled back the curtain on SearchGPT. 

SearchGPT is exactly what it sounds like. It’s a search engine infused with a healthy dose of the company’s GPT AI technology. It’s billed as a potentially smarter, context-aware alternative to the long-dominant Google Search. When you first boot it up, you’ll be greeted with a large text box asking, “What are you looking for?” Once you give it an answer, the app will get to work searching, but rather than returning a list of results similar to Google, SearchGPT will try to organize, summarize, and make sense of them. 

In one example, a user searches for music festivals in California during the announcement video. Once the user hits enter, SearchGPT scans the web similar to how ChatGPT does and pulls in the closest music festivals, gives each one a quick summary of what to expect and follows them up with affiliate links.

The company has stressed that SearchGPT is still only a prototype and is not quite ready for mass adoption. It’s only opening to 10,000 test users at launch. 

Earlier this year, Meta hinted at a significant development in AI: an open-source model that could rival the best from companies like OpenAI. Last week, Meta officially launched Llama 3.1, claiming it surpasses other leading models such as GPT-4o and Anthropic’s Claude 3.5 Sonnet in several benchmarks.

Llama 3.1, the largest open-source AI model to date, boasts 405 billion parameters and has shown superior performance compared to its competitors. The Llama-based Meta AI assistant is also expanding to more countries and languages and now includes a feature for generating images based on specific likenesses. CEO Mark Zuckerberg predicts Meta AI will become the most widely used assistant by the end of the year, overtaking ChatGPT.

Picture this: You’re running an e-commerce, marketplace, or content site and you get a new tool. All have to do is plug your data in, and within minutes, that tool will start fine-tuning a multi-modal neural model that shapes your raw data into tabular, language, image, or video embeddings. Within hours, you’ve got a ready model that you can use to run highly relevant recommendations and search features on your site or in your app.

That tool is called Shaped. The team behind Shaped says it’s making the same personalized AI that’s powering big tech (think "For You" feeds, recommendations, and search) accessible to any business. Shaped can update recommendation suggestions in real-time and its declarative SQL API makes it easier to update than competitor Recommend tools.

📱 Slack has announced a suite of new work-based widgets for iOS.

🌦️ Google Researchers built a new weather prediction model using AI.

📱 Apple is working on a foldable phone that could be released by 2026.

🤖 Google has made its Gemini chatbot faster and more generally available.

🔍 Microsoft has unveiled its latest AI feature called Bing Generative Search.

📺 Apple plans to debut Apple Intelligence a few weeks after the iOS18 launch.

🚶 Wiz, the cybersecurity start-up, has walked away from Google’s $23B offer.

💬 Telegram is planning to launch its very own app store after it hit 950M users.

🗺️ Apple Maps has launched on the web in a bid to compete with Google Maps.

✍️ Substack has updated its iOS app to allow both drafting and publishing posts.

What did you think of today's newsletter?

😻 Loved it.

😾 You could do better.