🚀Today’s Top AI Headlines:

1. Google Labs upgrades AI tools with new tricks: Google Labs has rolled out major updates to two of its experimental AI tools, Mixboard and Doppl, pushing them further into creative and consumer-focused use cases. Mixboard, Google’s AI-powered concepting and brainstorming board, now includes support for Nano Banana Pro, a lightweight model designed for rapid ideation and on-the-fly presentation generation. The upgrade enables users to explore ideas more fluidly, remix concepts instantly, and prototype slide decks without relying on heavy design workflows. Doppl, Google’s AI-driven fashion app, also received new enhancements. It now allows users to upload personal photos and generate a digital model of themselves to virtually try on outfits. Unlike typical AR try-on tools, Doppl uses AI to simulate how fabric drapes, how colors match skin tones, and how outfits move with the body, making it more realistic and personal. This update positions Doppl as a potential breakthrough in online fashion retail and consumer personalization. Both tools remain in limited rollout, with availability depending on region and access to Google Labs experiments. While still early-stage, these updates highlight Google’s broader strategy: blending creativity, personalization, and consumer applications into its AI ecosystem, giving everyday users advanced generative capabilities.

   Source: Google

   🤖 Robi: “If my avatar wears Crocs, just know it wasn’t my choice.”

2. ChatGPT now delivers groceries via Instacart: OpenAI and Instacart have expanded their long-running partnership with a major new feature: users can now shop for groceries directly inside ChatGPT without switching apps. This integration brings product discovery, cart building, and checkout into a single conversational flow, turning ChatGPT into a full-service shopping interface. With this update, users can ask ChatGPT to plan meals, generate recipes, or budget weekly groceries, and then instantly populate an Instacart cart with recommended items. ChatGPT can compare prices, swap alternatives, and tailor orders to dietary goals or household size. This puts AI into a more action-oriented commerce role, where it not only suggests but executes real-world tasks. The collaboration strengthens Instacart’s push to differentiate itself from grocery competitors by embedding AI into the shopping journey. For OpenAI, the move reflects its strategy to expand ChatGPT beyond information retrieval and into transactional use cases, where AI becomes a productivity layer for daily life. The integration is available across supported Instacart markets and works on both desktop and mobile versions of ChatGPT. As conversational commerce accelerates, this partnership signals a future where shopping becomes increasingly automated, personalized, and embedded in AI assistants.
   Source: OpenAI

   🤖 Robi: “Finally, a way to impulse-buy oat milk while debugging code.

   ”

3. Mistral drops blazing-fast open-source coding model: Mistral has unveiled Devstral 2, the next-generation model in its coding-focused lineup, along with Vibe CLI, its first autonomous coding agent designed to operate directly from the terminal. The headline model, Devstral 2 (123B parameters), delivers a 72.2% score on SWE-bench Verified, bringing it close to the top open-weight model, DeepSeek V3.2, despite being 5× smaller in size. The smaller variant, Devstral Small 2 (24B), stands out for its ability to run on a single GPU or even a laptop CPU, making high-level coding assistance more accessible to individual developers Mistral’s licensing approach is notable: Devstral 2 uses a modified MIT license restricting use by companies earning more than $20M per month, a move aimed at preventing the biggest players from absorbing its innovations cost-free. Meanwhile, Vibe CLI, released under an Apache 2.0 license, acts as a terminal-native coding agent capable of scanning entire codebases, proposing fixes, and performing multi-file updates autonomously. This dual release strengthens Mistral’s push into agentic coding workflows. By targeting both enterprise-scale engineering tasks and individual developer needs, Mistral positions Devstral 2 as a serious contender in the rapidly evolving ecosystem of autonomous software development tools.
   Source: Microsoft

   🤖 Robi: “Great, now even the terminal has a personality. What’s next, sarcastic semicolons?”

🔍Beyond the Headlines:

1. MIT teaches robots furniture assembly via voice: MIT researchers have created a groundbreaking “speech-to-reality” system that allows users to verbally instruct a robot to build physical objects, including furniture, without any modeling or robotics expertise. The system translates natural language into modular assembly steps, enabling users to disassemble and reconfigure parts into entirely new objects on demand. Beyond accessibility, the approach could drastically reduce manufacturing waste by extending the lifecycle of modular components. The project represents a major shift toward intuitive human–robot collaboration and hints at a future where spoken instructions directly shape real-world fabrication.
   Source: MIT

   🤖 Robi: “Great. Now the robot understands “screw that in backwards” better than I do.’’

2. AI finds hidden geothermal hotspot in Nevadae: Geothermal startup Zanskar has identified a previously hidden underground heat reservoir in Nevada using AI, potentially the first major industry-led geothermal discovery in decades. The system is classified as a “blind” geothermal reservoir, meaning it shows no surface indicators and is extremely difficult to detect using traditional surveying. Recent drilling confirmed temperatures suitable for power generation, suggesting the site could support a future geothermal plant. Scientists say this discovery indicates many more hidden reservoirs may exist across the western U.S., offering significant potential for clean, always-on renewable energy.

   Source: Wired

   🤖Robi: “Finally, AI doing something hotter than making slideshow templates.”

🤖Prompt of the Day:

Board Governance Optimization Framework

🤖AI Tools You Didn’t Know You Needed:

Problem: Typing, especially long docs, code, or multilingual content, is slow and error-prone when done manually.

AI Solution: LazyTyper uses fast, accurate speech-to-text so you can “talk instead of type,” producing content 3× faster with high accuracy.

AI Tool: LazyTyper is a voice-typing assistant powered by 12 AI speech models (including offline/local options) that turns spoken words into clean, structured text for documents, coding, notes, and emails.

Helpful Features

• Multi-Engine Speech Models: 12 AI voice models, including 5 fully offline.

• Fast & Accurate: Type up to 3× faster with ~90% accuracy.

• Multi-Language Support: Dictate in various languages for writing or coding.

• No Ads, No Limits: Completely free with full feature access.