ICYMI: Google unveils visual AI mode update with Gemini 2.5 in US

Google's updated AI Mode in Search uses its Gemini 2.5 model to deliver visual and conversational results, combining natural language and image input for refined searches. It supports real-time product discovery through the Shopping Graph.

🗞 #gemini

Microsoft announces Agent Framework preview

Microsoft's Agent Framework enters public preview, offering developers and enterprises a unified SDK and runtime for building and governing scalable multi-agent AI systems, with strong observability, open standards, and enterprise integration.

🗞 #microsoftapps

Hume AI launches Octave 2 and EVI 4 mini voice models

Hume AI released Octave 2, a faster, multilingual voice AI model supporting 11 languages and optimized for advanced hardware. It cuts latency to under 200ms, reduces costs by 50%, and introduces new features for dubbing and voice editing applications.

🗞 #hume

Anthropic launches Claude integration for Slack workspaces

Anthropic’s Claude AI is now integrated into Slack for paid Team and Enterprise users, offering document analysis, drafting, and contextual replies by accessing messages, files, and connected tools, all within Slack’s security framework.

🗞 #claude

Cursor rolls out hooks, team rules and sandboxed terminals

Cursor 1.7 introduces Hooks for run-time Agent customization, team-wide rules for policy standardization, sandboxed terminals for secure execution, deeplink sharing, and expanded Agent monitoring—all tailored for secure team collaboration.

🗞 #cursor

10000 Sora 2 invites to claim 👀

UPD: All invites were claimed. However, someone build a website from where you can snipe an invite as well https://formbiz.biz

OpenAI works on user profiles and direct messaging for ChatGPT

ChatGPT Android beta adds early implementation of profiles and direct messaging. These updates indicate a shift toward building a social platform around OpenAI's ecosystem.

🗞 #chatgpt

Comet browser is getting a Split View! 👀

Google released Nano Banana model in GA (non-preview) with support for a broader range of different aspect ratios. Already available on AI Studio.

Supported ratios include:
- Landscape: 21:9, 16:9, 4:3, 3:2
- Square: 1:1
- Portrait: 9:16, 3:4, 2:3
- Flexible: 5:4, 4:5

Google launches CLI tool for Jules SWE Agent

Jules introduces a CLI tool that integrates its platform into the terminal, allowing developers to run tasks and chat from local environments. It supports parallel workflows, aligning with development routines beyond the browser.

🗞 #jules

NotebookLM will get quick filters on the Sources tab so you can view them for each type: docs, links, pasted text, YouTube, etc.

Google launches Gemini 2.5 Flash Image model in general availability

Google has released Gemini 2.5 Flash Image for global use via Gemini API, AI Studio, and Vertex AI, offering scalable image generation with support for 10 aspect ratios, multi-angle rendering, character consistency, and natural language editing.

🗞 #aistudio

Personalisation features are rolling out on NotebookLM for chat conversations!

Define your conversational style
- Default
- Learning Guide
- Custom

Sora 2 Pro is rolling out to Pro users 👀

It comes with HD videos and up to 15s long duration

NotebookLM Video Overviews will get new styles, including Anime

Google is expanding NotebookLM's video overview feature with new visual styles and a format selector, signaling broader customization options. These updates align with upcoming AI model improvements and a larger product rollout strategy.

🗞 #notebooklm