Welcome back!

Google launched Gemini 2.0, a leap forward in multimodal AI with real-world applications. Featuring native AI agents designed for web navigation, coding, and research, this flagship release offers a glimpse into the future of AI. Let’s unpack…

Today’s Summary:

• 🚀 Google launches Gemini 2.0 with built-in AI agents

• 🍎 Apple integrates ChatGPT into iOS 18.2

• 🔧 Jules AI codes while you sleep

• ✍️ OpenAI expands Canvas access to free users

• 💬 YouTube debuts AI auto-dubbing for videos

• 🤖 Devin coding assistant goes live

• 🛠️ 2 new tools

TOP STORY

Google launches Gemini 2.0 with native AI agents

The Summary: Google has unveiled Gemini 2.0, a flagship AI designed to act as an agent on users' behalf. The model handles text, images, audio, and video, it also natively produces images and multilingual speech. The initial release includes three specialized AI agents for web browsing, coding, and universal assistance. Gemini Flash 2.0 runs twice as fast as its predecessor. Integration into Google products begins today, with broader adoption planned for 2025.

Key details:

• Gemini 2.0 Flash is 2x faster than Gemini 1.5 Pro with same performance

• Reports 92.9% accuracy in coding and 89.7% in math tasks. Beats o1 full and Claude Sonnet 3.5 in SWE-Bench.

• Deep Research feature processes up to 1 million tokens to generate reports

• Project Astra a universal multimodal assistant, includes Google Maps integration and a 10-minute context memory

• Project Mariner navigates websites automatically, requiring human approval for sensitive actions like purchases

• Native image generation and multilingual text-to-speech are set to arrive by January 2025

Why it matters: Gemini 2.0 is a major release moving AI from passive assistance to active task completion, with agents capable of researching, coding, and navigating the web autonomously. The emphasis on real-world applications suggests a practical approach to AI advancement that balances capabilities with human control.

Try Gemini 2.0 Flash in Gemini App. Select “Gemini 2.0 Flash Experimental” in the dropdown menu.

APPLE

Apple adds ChatGPT to iPhones with iOS 18.2 update

The Summary: Apple integrated ChatGPT across iPhones, iPads, and Macs with the iOS 18.2 and Sequoia 15.2 update. The feature brings OpenAI capabilities into Apple's ecosystem while prioritizing user privacy. ChatGPT access is optional, and user data remains protected by default. Alongside ChatGPT, the update includes visual intelligence for camera and creative tools like Image Playground and Genmoji.

Key details:

• Requires iPhone 16+ or iPhone 15 Pro/Pro Max, or Mac M1+ for on-device AI processing

• Visual intelligence lets users point their camera at objects for instant information, using ChatGPT or Google Search

• ChatGPT can be used without an account, and OpenAI won't store requests or use data for training in this mode

• Adds localized English support to 6 new regions including Australia, Canada and the UK, with 11 more languages planned

Why it matters: Apple's ChatGPT integration offers a privacy-first approach to AI services integration. Users with existing OpenAI accounts can get additional features and device continuity with increased privacy.

DEEPMIND

Jules writes code while developers sleep

The Summary: Google has unveiled Jules, an experimental AI coding tool powered by Gemini 2.0. Jules autonomously fixes bugs, implements code changes, and prepares detailede pull requests, integrating with GitHub workflows. The system maintains human oversight by requiring explicit approval before merging code changes.

Key details:

• Poor code quality costs industry $2.84 trillion annually according to CISQ

• Lawrence Berkeley Lab researchers reduced analysis time from 1 week to 5 minutes using Jules and related tools

• McKinsey reports large IT projects routinely exceed budgets by 45% and deliver 56% less value than predicted

• Gartner predicts AI-assisted coding will be involved in 75% of new application development by 2028

Why it matters: Jules moves from suggestion-based software development tools to autonomous coding agents. This prototype offers a glimpse into the future of AI-driven coding tools, which could take over routine maintenance tasks, allowing developers to focus on orchestrating complex systems and driving innovation.

Join the waitlist

QUICK NEWS

Quick news

• OpenAI rolls out Canvas writing tool interface to free users

• Devin coding assistant is generally available

• YouTube launches AI auto-dubbing tool

TOOLS

🥇 New tools

• Remento - The AI biographer for loved ones

• Mentor - Give your goals a head start