Good morning. It’s Friday, June 27th.

On this day in tech history: in 2016, Google DeepMind dropped new details on AlphaGo’s reinforcement learning tech around June 27. Fresh off its Lee Sedol beatdown, the system showed AI could crush complex games, pushing the needle on decision-making algorithms.

In today’s email:

• Google’s Major AI Releases This Week

• Claude’s No-Code App Builder

• 5 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

_{In partnership with Adaline}

🚀 Build Smarter, Ship Faster with Adaline

Adaline is the end-to-end platform trusted by world-class product and engineering teams to iterate, evaluate, deploy, and monitor large language model applications—all in one place.

🛠 Prompt like a pro: Test across datasets, compare models, and collaborate seamlessly—with automatic versioning and prompt management that actually works.

⚙️ Deploy without drama: Adaline powers 200M+ API calls a day with 99.998% uptime, handling scale effortlessly and securely.

📈 From idea to insight: Move from sketch to live deployment in record time, with real-time logs, analytics, and performance monitoring.

💡 Ready to launch? Adaline is now generally available—with $1M in API credits for new users.

_{Thank you for supporting our sponsors!}

Google’s Major AI Releases This Week

Gemini CLI drops into the terminal, letting devs write, debug, and refactor code using natural language, no IDE required. It's fast, open source (Apache 2.0), and hooks into tools like Veo 3 for video, Google Search for real-time data, and Deep Research for extended tasks. High limits (60 req/min, 1,000/day) mean it’s ready for serious usage, though trust in AI-generated code still lags.

Colab AI levels up Colab notebooks with a Gemini assistant that runs analysis workflows, trains models, and cleans data, all inline. A new Data Science Agent even iterates on pipelines based on user feedback. No more tab-hopping or copying code between cells.

Imagen 4 sharpens AI visuals with better prompt accuracy and pro-level text rendering. Now live in Gemini API and AI Studio, it comes in standard ($0.04/image) and Ultra ($0.06/image) tiers. Ultra’s free to try for now, tuned for high-fidelity image output.

Gemma 3n brings AI to edge devices. This open-source model (E2B/E4B) is optimized for low memory (as little as 2GB RAM), thanks to efficient design using MatFormer and MobileNet-V5. It handles multimodal input and outputs text, performing above its weight class, E4B even breaks 1300 on LMArena.

Doppl, a new app under Google Labs, lets users virtually try on clothes using a full-body photo. It pulls garments from stores, friends, or social feeds and creates shareable try-on videos. Accuracy may vary.

Claude Just Became a No-Code AI App Engine

Anthropic just upgraded Claude from chatbot to full-on no-code platform. Users can now build and share functional AI apps called artifacts, directly inside Claude. These aren’t static outputs; they’re live tools that take input, adapt on the fly, and run without needing a conversation in progress. Claude now embeds intelligence directly into these tools, eliminating copy-paste workflows and turning prompts into working software.

Since launch, users have created over 500 million artifacts: flashcard generators, spreadsheet analyzers, smart tutors, and NPC-driven games. Free users can build and interact; paid tiers unlock more power. Anthropic handles content moderation with multi-layered safeguards. As OpenAI pushes GPT-powered agents, Anthropic bets on apps with real UX. Read more.

• This AI-powered startup studio plans to launch 100,000 companies a year — really

• Meta Escalates AI Offensive with Talent Raids, Gets OpenAI's Top Minds

• OpenAI Pushes API and DevDay 2025 as Microsoft Blocks IPO Plans

• Murati’s AI Play: Ditch AGI, Build Tools That Work

• AlphaGenome by Google Analyzes DNA Mutations and Maps Their Biological ImpactIn just 3 months, CoreWeave CEO, once a crypto-mining bro, becomes a deca-billionaire

• DeepSeek's newest model delayed by GPU export restrictions

• Watch: Humanoid robots face-off ahead of China's first-ever 3-on-3 AI football match

• AI is doing up to 50% of the work at Salesforce, CEO Marc Benioff says

• Xiaomi Debuts AI Glasses with 2K Video, Voice Assistant for 1,999 Yuan

• Human-AI "Couples Retreat" Exposes Emotional Limits of Virtual Companions

• Attorneys Weigh In: AI Is Already Altering Legal Workflows

• Tesla’s Robotaxi Pilot in Austin Faces Hiccups With Erratic Driving, Rider Complaints

• Sesame snaps up another Oculus VR co-founder

• Agentic AI deployment accelerates despite risks: KPMG

• Watch: ElevenLabs debuts Voice Design v3 with ultra-custom AI voices

• IBM pushes multi-LLM strategy as enterprises reject one-size-fits-all AI

• For Replit’s CEO, the future of software is ‘agents all the way down’

• Creative Commons debuts CC signals, a framework for an open AI ecosystem

• Berkeley unveils customizable humanoid robot built from 3D-printed parts

• This AI system makes human tutors better at teaching children math

• MIT, NUS Build Memory-Savvy Framework for Smarter Language AI

• AI avatars are here in full force—and they’re serving some of the world’s biggest companies

📄 ShareGPT-4o-Image: Aligning Multimodal Models with GPT-4o-Level Image Generation

📄 Matrix-Game: Interactive World Foundation Model ​​

📄 JarvisArt: Liberating Human Artistic Creativity via an Intelligent Photo Retouching Agent 

📄 MMSearch-R1: Incentivizing LMMs to Search

📄 AnimaX: Animating the Inanimate in 3D with Joint Video-Pose Diffusion Models

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, reply to this email or DM us on 𝕏!