Good morning, {{ first_name | AI enthusiasts }}. Apple's answer to falling behind in AI? Build an army of home robots, complete with motorized arms, AI personalities, and a long-awaited Siri upgrade built from the ground up.

But with an Apple Intelligence emergency already on hand, can Apple's hardware talents overcome its internal AI struggles to crack the code on the next big home device?

In today’s AI rundown:

Apple plots AI comeback with home robots

OpenAI brings back 4o after GPT-5 anger

Create a self-updating slide deck with Manus

Microsoft goes on the offensive for Meta AI talent

4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

APPLE

🏠 Apple plots AI comeback with home robots

Image source: Apple ML Research

The Rundown: Apple is pushing for a smart home push with four AI-powered devices, according to a new report from Bloomberg, including a desktop robot, smart display, and security cameras — targeting launches between 2026 and 2027.

The details:

The desktop robot will feature a motorized arm that controls a display to track users and lock onto speakers, infused with an AI-upgraded Siri.

Bloomberg said Apple is planning a personality-driven character for the new Siri called “Bubbles,” comparing it to Clippy from the original Microsoft Office.

A smart display is also set to arrive mid-2026, running a new ‘Charismatic’ OS for controlling home apps, music, web browsing, and more via voice or touch.

New AI-powered security cameras with months-long battery life will automate household tasks like dimming lights or playing personalized music.

Engineers are rebuilding Siri from scratch with AI models under the codename Linwood, while testing Anthropic's Claude as a backup codenamed Glenwood.

Why it matters: Apple’s much-hyped AI-powered Siri overhaul is still just talk, and the home AI robotics market has yet to see a product that truly sets the standard. With Google, Amazon, and OpenAI’s Jony Ive–designed device advancing quickly, Apple has little time to make its move.

TOGETHER WITH SANA

💡 The most powerful way to bring AI to your company

The Rundown: Most teams have tried AI tools, but connecting them to your business context changes everything. Sana Agents is an all-in-one platform that gives you no-code AI agents grounded in your company data. Deploy across teams in minutes, not months.

With Sana Agents, you can:

Draft reports, generate presentations, analyze data, and handle emails through agents

Connect to 100+ platforms, including Salesforce, Teams, and your CRM

Automate workflows that know your processes and permissions

Scale from isolated AI experiments to real work impact

Drive enterprise-wide innovation today with Sana Agents.

OPENAI

🔄 OpenAI brings back 4o after GPT-5 anger

Image source: ChatGPT

The Rundown: OpenAI CEO Sam Altman announced a series of changes to ChatGPT following backlash from the company’s GPT-5 launch, including the return of the popular 4o model, expanded rate limits, and new controls for model choice.

The details:

GPT-4o is returning to the model picker for all paid users, with Altman saying there will be “plenty of notice” if the model is ever deprecated.

Weekly rate limits for advanced reasoning in GPT-5 jumped from 200 to 3,000 queries, with Altman also clarifying the 196k context window for the new model.

Users also gain new "Auto," "Fast," or "Thinking" options for GPT-5, addressing anger from queries frequently being routed to the wrong model at launch.

Altman also revealed a personality update is coming for GPT-5, but said the real learning is the need for “per-user customization and model personality.”

Why it matters: GPT-5 is by nearly every measure a strong step forward, but a rocky rollout and forced user actions set a bad tone for what was a massively hyped launch. The 4o saga also shone the light on a (clearly larger than anticipated) corner of the user base that cares more about personality than coding or benchmarks.

AI TRAINING

🤖 Create a self-updating slide deck with Manus

The Rundown: In this tutorial, you’ll learn how to use Manus, an AI “super agent” that can handle complex, multi-step tasks, to build professional and data-rich slide decks automatically — without the manual work.

Step-by-step:

Give Manus a clear, detailed prompt in the dashboard (e.g., “10 AI opportunities for my hospitality business + 1 slide on near-term trends”)

Select Slides as your output, choose a theme, and click Confirm to start the process

Let Manus work in the cloud — it will research, gather data, and design your slides with visualizations like graphs while you do other things

When done, review the deck, export to Google Slides for edits, or go back to set up automatic updates for fresh content

Pro tip: Include audience, purpose, and format in your prompt to get a polished, ready-to-present slide deck.

PRESENTED BY IBM

🚀 Lockheed Martin partners with IBM for AI

The Rundown: Lockheed Martin’s AI takeoff began with data cleanup — cutting its tools by 50% and replacing them with a single unified system that runs on IBM’s watsonx.data to drive faster, smarter engineering.

Learn how Lockheed Martin leveraged watsonx to:

Help 10,00 engineers build scalable AI products in its AI Factory

Respond to employee questions using AI, boosting accuracy by 20%

Overcome data silos to create a single accessible, connected data environment

Read the full story. 

AI TALENT WARS

🎣 Microsoft goes on the offensive for Meta AI talent

Image source: gpt-image / The Rundown

The Rundown: Microsoft is targeting Meta’s AI talent in a new recruiting offensive, according to a report from Business Insider, using multi-million dollar offers of its own to lure researchers from labs outside of the new Superintelligence Labs division.

The details:

Microsoft is reportedly aiming to match Meta’s compensation and using ‘special recruiting teams’, with a list of targets circulating via hiring managers.

Teams targeted include Reality Labs, GenAI Infra, and Meta AI Research, with recruiting led by Mustafa Suleyman and former Meta engineer Jay Parikh.

New processes for “critical AI talent” allow for streamlined offers and higher-up approvals within 24 hours.

Why it matters: Microsoft is taking a page out of Meta’s own playbook, though matching the money that Zuck has shown the willingness to give to top AI talent will be no small feat. That said, with reports of Meta’s AI unit being plagued by culture issues, it’s possible that some of the non-MSL employees are feeling ready for a move.

QUICK HITS

🛠️ Trending AI Tools

🎥 Pika Labs - New audio-driven performance model for HD quality videos

🤖 Mistral Medium 3.1 - Upgraded model with performance improvements

👀 Hunyuan-Large-Vision - Tencent’s new multimodal understanding model

🤝 Cobot - AI teammate for automating tools you already use

📰 Everything else in AI today

Igor Babuschkin announced is leaving xAI, starting Babuschkin Ventures to invest in AI startups that “ advance humanity and unlock the mysteries of our universe.”

Anthropic is acquiring three co-founders and several team members of Humanloop, an enterprise AI evaluation and safety platform.

The United States is reportedly secretly placing tracking devices in shipments of advanced AI chips from Nvidia and AMD to identify potential reroutings to China.

Tencent released Hunyuan-Vision-Large, a multimodal understanding model that slots in at No. 6 in the Vision Arena leaderboard, near GPT-4.5, o4 mini, and 4 Sonnet.

Google announced the rollout of several new features for Gemini, including temporary chats and memory to reference previous conversations and learn user preferences.

Higgsfield AI launched Draw-to-Video, allowing users to sketch text directions, shapes, and visual instructions on images to create a tailored video output.

‘Godfather of AI’ Geoffrey Hinton proposed training “maternal instincts” towards humans into AI as a potential solution to preventing the tech from wiping out humanity.

Liquid AI introduced LFM2-VL, open-weight vision language models designed for fast performance on consumer devices.

COMMUNITY

🤝 Community AI workflows

Every newsletter, we showcase how a reader is using AI to work smarter, save time, or make life easier.

Today’s workflow comes from reader Raj P. in Toronto, Canada:

"I use Perplexity AI Labs for data analytics. As a data scientist, I employ data science tools to discover trends and insights for themes ranging from business to lifestyle. I’ve developed a scaffold to discover hard-to-reach insights. People are adopting healthy lifestyles, for example, adjusting temperatures to 60s at night for deep sleep and a healthy body. It’s a handy hack, but super fun as well!"

How are you using AI in your work? Tell us here.

🎓 Highlights: News, Guides & Events

Read our previous newsletter: Silicon Valley's messiest feud explodes

Today’s AI tool guide: Run advanced AI models on your computer

Watch our last workshop: AI agents for Customer Service

See you soon,

Rowan, Joey, Zach, Shubham, and Jennifer—the humans behind The Rundown