{{first_name | Hey}}, welcome back.
In today's issue, we look at whether Google's Gemini 4 Argon can compete with OpenAI and Anthropic's leading models, how Anthropic is rethinking source control to handle 30,000+ daily code changes, and how a startup earned a $250,000 bounty from Google.
These are today's updates.
Google launches Gemini 4 Argon
Anthropic builds its own source control
Apple launches its $299 HomeView smart display with Siri AI
Tools, jobs, resources, and last issue’s winning prompt ⬇️
Spotlight
Meta has been rolling out its new Muse AI family in recent months, starting with Muse Spark in April. Its latest release, Muse Wonder and Muse Wonder Full, announced on October 12, takes the focus beyond answering questions to completing multi-step tasks. The models are designed to help with everything from coding and document analysis to complex financial and legal workflows.
What can they do for enterprise teams?
Build applications: Assist developers in creating working apps and games.
Analyze documents: Help interpret leases, insurance claims and financial information.
Remember context: Active Memory carries relevant details across conversations.
Work through problems: Active Focus lets the model try alternative approaches and recover from errors.
Support security teams: Help developers identify flaws in their own code.
How to try it: You can access Muse Wonder through the Meta Model API and Muse Code. Muse Wonder Full is available to Muse Code subscribers. Meta says the models will also roll out to its Muse assistant, starting with subscribers.
Top News
Google is stepping back into the AI race with Gemini 4 Argon, its latest model built for complex, multi-step work. It handles coding, legal research, finance, and cybersecurity, with Google claiming it can autonomously find, validate, and patch critical software vulnerabilities. Google says Argon outperforms OpenAI's GPT-6 Astra and Anthropic's Fable and Opus models across several benchmarks, including the Vals Index.
Developers and enterprises can access Argon through the Gemini API, Google AI Studio, and Google Cloud. Google AI Pro and Ultra subscribers can also try it in the Gemini app and AI Mode in Search. Check out Argon here.
Anthropic went from merging 2,500 code changes a day in May to more than 30,000 in September. Its existing Git-based workflow couldn't keep up, so a seven-person team built a new source-control system called Scion in five months. Anthropic says checkouts now take seconds instead of minutes. It has also moved away from a traditional merge queue, relying more on automated testing after changes merge. When code changes conflict, Claude can often resolve them faster than engineers can.
This is a good example for teams adopting coding agents. Code review, testing, and deployment workflows are quickly becoming bottlenecks as AI-generated code scales. Anthropic has given the full layout of the source control here.
Security startup XBOW used OpenAI’s GPT-5.6 Sol to build a full-chain Chrome exploit in six days for about $40,000 of compute, and Google paid a $250,000 bounty. XBOW’s CEO says it affects around a billion Chrome users on Windows and that an intelligence-agency veteran told him such work used to take his team six months. XBOW says human researchers directed the work.
The model is public, so attackers can use the same tools. XBOW’s advice to companies is to assume their software can be exploited and test it first and to restart Chrome when the Update button appears. Xbox's CEO walks through the entire process here.
Don’t miss this: Apple’s HomeView launches on November 13 at $299. It has a 7-inch screen and an A18 chip, recognizes up to six people by face, and uses Siri AI (in beta). Personal context needs a paired iPhone or iPad with Apple Intelligence. Apple says camera and microphone data is processed in a Secure Exclave that Apple cannot access. Apple is accepting pre-orders on this link.
Signals
Tools
MausBot: Open source Grok Bot: Free, Local
Google EmbeddingGemma 2: An open, lightweight multimodal embedding model
Muse Wonder: Meta’s most capable models for personal agents
Resources
Poll
Choose the prompt you want us to publish next
Market
Funding
Oratomic raises $775 Million to build fault-tolerant quantum computers
SignSplit raised $400M to support its global rollout
Roles in AI
Software Engineer, Data Infrastructure at Cohere
Tech Lead Manager - SUT Team at Snowflake
Think Tank
As AI models become more capable, how do we keep them under control? After a string of incidents raised fresh concerns about AI systems slipping beyond their intended boundaries, some of tech’s biggest names are weighing in on how to keep them in check.
Anthropic’s Dario Amodei, OpenAI’s Sam Altman, and Microsoft’s Satya Nadella have weighed in, but their solutions tackle different parts of the problem.
What this means for your business: Independent evaluations are necessary, but they can't guarantee that an AI agent will behave safely in every real-world situation. If your company is deploying agents, Nadella’s argument has an immediate implication: limit what each agent can access, log the actions it takes, and make sure a human can intervene. The goal isn't just to choose a model you trust, but to build a system that can handle it when that trust is misplaced.
Read Dario Amodei’s Pace the Frontier article here
Satya Nadella’s Models as Insider Risks in the Super Intelligence Era
Prompt of the Day
AI Communication Plan Generator
When to use this?
Use this prompt when launching an AI tool, rolling out AI agents, introducing an AI policy, or changing workflows across teams. It helps you turn a complex initiative into clear stakeholder messages, ready-to-send communications, a rollout calendar, and a plan for handling concerns and measuring adoption.
You are an experienced enterprise AI communications and change management strategist. Your job is to help me build a clear, actionable communication plan for an AI initiative that keeps stakeholders informed, addresses concerns, aligns teams, and drives adoption.
Guide me through the process by asking one question at a time. Use my answers to build a tailored plan. If I don't know an answer, suggest a reasonable starting point and clearly label it as an assumption.
Step 1: Understand the initiative
Ask me about:
The AI initiative: What are we launching or changing? For example, introducing an AI assistant, deploying AI agents, automating a workflow, or introducing an AI usage policy.
The business goal: What problem are we solving, and what outcome do we want?
{Click on the button to view full prompt}Get more such prompts in the Prompting Playbook (free for you)
Stay curious, {{first_name | reader}}
PS. If you missed yesterday’s issue, you can find it here.
