Tech • AI • Robotics • Game
So Much AI News: New ChatGPT Tools, NotebookLM Updates, Big Claude Changes, Grok Bot Live, + More!
9.2/10TL;DR
Anthropic, Google, OpenAI, xAI and Meta rolled out a wave of AI product updates centered on simpler interfaces, stronger voice tools, collaborative work features and low-cost agent access.
KEY POINTS
Claude unifies chat and cowork
Anthropic is merging Claude Chat and Claude Cowork into a single experience that automatically decides whether a prompt should be handled as conversation or task-oriented work. Existing setups such as projects are expected to remain intact, while the change removes a layer of manual switching that had made the interface feel fragmented.
New Claude outputs expand document and design work
A new Output menu in Claude adds Docs, Slides and Design generation directly inside the main interface. Slides can be edited in place and exported as PDF, web page or PowerPoint, while Docs are built section by section with live editing on the side. The integrated Design tool produces polished motion graphics and web-style outputs, though some users may find that previously created design systems are not yet fully accessible in the new workflow.
Task delegation arrives in Claude Code
Anthropic also began rolling out a feature in Claude Code that lets one main interaction spawn additional tasks to handle delegated work. The setup effectively turns a single Claude session into a coordinator that can distribute subtasks, hinting at a broader push toward chief-of-staff style agent behavior across Claude products.
NotebookLM gains recording and live conversation
Google is preparing two notable upgrades for NotebookLM. Users will be able to record audio directly into a notebook from a phone rather than uploading existing files, and they will also be able to hold a live voice conversation with notebook content. The additions are aimed at students, meeting-heavy workers and anyone collecting ideas on the go, with rollout expected over the next few weeks.
Gemini live voice gets more natural
Google introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking, both focused on real-time voice interaction. The models are designed to better handle interruptions, switch languages more smoothly and interpret visual context more accurately, while the extended-thinking version can reason and speak in parallel for a more seamless back-and-forth. Early hands-on impressions suggest a clear step forward, even if ChatGPT still feels smoother to some users.
Google tests a family coordination agent
Google’s CC experiment is shifting from a personal morning summary into a family-focused assistant. It can connect up to five members, combine family calendars into a shared daily briefing, create shared Google Calendar and Google Tasks items, and coordinate through Google Chat. The tool is positioned as a practical way to reduce scheduling gaps inside households.
More Google ecosystem integrations land
Google Flow is now available on iOS, making its image and video generation tools easier to use from a phone. Gemini is also adding more connected apps, including Whisper integration for bringing notes from an external note-taking workflow into Gemini.
OpenAI expands website building and desktop controls
OpenAI upgraded ChatGPT Sites, its website creation and deployment tool, with team collaboration, private sharing and custom domains. The company also updated ChatGPT Pets, adding quick access to new chats and live voice, plus a lighter Mini mode that keeps task notifications visible while using less screen space.
ChatGPT moves deeper into Microsoft Office
ChatGPT is being added to Microsoft Word, Excel and PowerPoint, where it appears in a side panel and can help edit documents, work with spreadsheets and assist on presentations. The integration also supports connected apps and custom skills, extending ChatGPT from standalone chat into everyday office workflows.
xAI and Meta push consumer agents
Grokbot is rolling out live voice on desktop and mobile, bringing it closer to competing assistants that already support real-time spoken interaction. Meta’s Muse is using aggressive incentives to build adoption, including a referral offer that grants 1 billion Muse tokens to both users when an invite code is entered within 48 hours of sign-up. Reports also point to a future phone-calling feature that could let Muse contact businesses and complete basic bookings.
CONCLUSION
This week’s updates show AI platforms converging on the same goal: fewer menus, more natural voice interaction and tools that can turn chat into real work. The competitive edge is increasingly shifting from raw model power to usability, collaboration and agent-style execution.
Ask a question