Bot Eat Brain
Posts
Anthropic, Palantir, and AWS announced a partnership.

Anthropic, Palantir, and AWS announced a partnership.

PLUS: How to use AI to gather data.

Michael Parrish
November 08, 2024

TOGETHER WITH FETCHFOX.

Good morning, human brains, and welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

Claude chose violence today. 🔒 🤖
Anthropic, Palantir, and AWS partnered up.
Microsoft's new boss isn't human 🤖 🚗
Microsoft revealed Magnetic-One.
Scrape like a hacker (no code) 🦊 🔎
Scrape any website with plain English.

New here? Subscribe! 😎

Want to sponsor Bot Eat Brain?

🌎 Reach: 23,000+ readers

📩 Open Rate: 40%+

📍 Location: 80% USA, Canada, & UK

Peep today's ‘What Would You Do?’ at the bottom. 👇

MAIN COURSE

Claude’s new gig = war crimes? 🔒 🤖

On Thursday, Anthropic announced a partnership with Palantir and AWS. The goal is to bring its Claude AI models to U.S. defense agencies.

Source: @PalantirTech

What's the deal?

Claude 3 and 3.5 will be integrated into Palantir's AI Platform, hosted on AWS's most secure government infrastructure (Impact Level 6).

What's Impact Level 6?

The highest security level for defense systems, reserved for critical national security data requiring maximum protection.

War. Huh. Yeah. What's it good for?

Defense officials can use Claude for rapid data analysis, pattern recognition, and decision-making in classified settings.

Who else is courting the Pentagon?

Meta recently opened Llama to defense partners, while OpenAI is seeking closer DoD collaboration.

Why is this significant?

It shows top AI companies actively pursuing defense contracts as AI becomes crucial for national security.

What are they saying?

Palantir's CTO calls it “the next generation of decision advantage” for critical missions.

SIDE SALAD

4 AI agents > Your entire staff 🤖 👔

On Wednesday, Microsoft revealed Magnetic-One. It’s a system that coordinates multiple AI agents to handle complex tasks together.

Source: Microsoft

How does it work?

An “Orchestrator” agent leads four specialized AIs, like a manager directing a team to tackle multistep challenges.

What can it do?

Everything from ordering sandwiches to analyzing stock trends — any task requiring multiple steps and different skills.

Is it locked down?

Nope. Microsoft made it open-source and included a testing tool called AutoGenBench.

How well does it perform?

Holds its own against top agent systems in benchmarks like GAIA, AssistantBench, and WebArena.

Why is this exciting?

It brings us closer to having AI teams that can handle your daily to-do list autonomously.

What's the bigger picture?

Multi-agent coordination could be key to making AI useful for complex real-world tasks.

COOL TOOL 🛠️

Digestible AI

Making AI simple, accessible, and digestible.

Digestible AI is a content ecosystem that simplifies complex AI trends into actionable insights, tutorials, and tools—perfect for professionals, engineers, and AI enthusiasts.

Stay informed and discover practical ways to apply AI in your world (no tech jargon required)

Subscribe for free.

YOUR DAILY MUNCH

Startup News 💰

Google accidentally released a preview of Jarvis AI. It’s an extension designed to autonomously perform tasks like shopping, booking flights, and more.

KLING is launched Custom Models. It allows you to create personalized video characters by training them with 10-30 video clips for consistent appearances.

Research 👨‍🔬

HtmlRAG — a new approach that uses HTML instead of plain text to provide richer information for Retrieval-Augmented Generation systems.

Ask and it shall be given — researchers have proven that prompting AI models is Turing-complete, meaning a single Transformer can perform any function with the right prompts.