A16Z’s oxygen for AI startups 💨

Here’s how they’re beating the giants

October 25, 2024

Welcome back! This week, Apple’s got some thoughts on reality—or at least, reality as seen through your iPhone camera. In a world where AI can turn a sunset into a sci-fi landscape or swap out a photobomber for a beach scene, Apple is playing the responsible friend role. Its new “Clean Up” feature lets you remove those pesky background objects but stops short of letting you create a “fantasy” version of your last vacation.

It’s all about keeping your photos real… or at least, real-ish.

Anthropic’s Claude 3.5 Sonnet Can Control Your Computer

AI News

Anthropic just took a leap forward with Claude 3.5 Sonnet, its latest AI model that can interact directly with computers like a human. The new “computer use” feature, available in public beta, lets the AI view screens, move cursors, click buttons, and even type text. 

Ahead of the curve: Claude 3.5 Sonnet sets Anthropic apart from competitors like Microsoft and Google, who have yet to widely release similar features. While Microsoft’s Copilot Vision and Google’s Gemini app have shown early signs of this direction, they haven’t matched the public accessibility of Claude’s capabilities.

  • For developers, this opens up a playground of possibilities—enabling automation that goes beyond simple commands, potentially revolutionizing how software interacts with human workflows.

Key features of Claude 3.5 Sonnet:

  • Direct interaction: By typing and moving cursors, the AI mimics human interactions in a digital interface.

  • API access: Developers can integrate this new capability through Anthropic’s API, making it possible to build AI solutions that automate a wide range of tasks.

  • Public beta: Unlike some competitors, Anthropic has made this feature available to a broad audience, aiming to gather feedback and refine the model through real-world usage.

  • Experimental stage: While promising, the feature is still prone to glitches, making user feedback critical for future improvements.

What it means for AI: Claude 3.5 Sonnet’s ability to interact with computers could open doors for more dynamic AI applications—from automating repetitive office tasks to assisting in software development. Imagine an AI that can troubleshoot issues by navigating software interfaces or handle customer support tickets by directly interacting with backend systems. 

It’s a glimpse into a future where AI not only generates ideas but executes them, transforming how we think about digital work and bridging the gap between human-like understanding and machine precision.

Andreessen Horowitz’s ‘Oxygen’ for AI computing

CNBC

Andreessen Horowitz (a16z) is giving AI startups a lifeline with its “Oxygen” GPU cluster, which offers access to thousands of Nvidia H100 GPUs. The goal? To help smaller players get the compute power they need without battling giants like Google, Meta, or Microsoft.

The existing problem: Securing GPU access has been a challenge for AI startups, with the market favoring large companies that can lock in long-term contracts. For emerging AI developers, this creates a bottleneck, limiting their ability to train models at scale. 

How Oxygen makes a difference:

  • Flexible access: Oxygen allows startups to rent GPUs without committing to lengthy contracts, making it easier to scale up or down based on their needs.

  • Cost-effective support: Startups can access cutting-edge hardware without massive upfront costs, making the technology more accessible to those with limited budgets.

The big picture: Initiatives like Oxygen help democratize access to high-powered computing. For startups, it’s an opportunity to innovate without being squeezed out by industry behemoths, and for a16z, it’s a way to attract new startups and solidify its role as a key player in the next generation of AI.

Zoom and Suki’s AI Scribe—a Boost for Telehealth

Fast Company

Zoom is expanding its role in telehealth through a new partnership with Suki, an AI-powered medical note-taking tool. This collaboration integrates Suki’s AI scribe into Zoom’s platform, helping doctors streamline their administrative workload during virtual consultations.

Zoom’s AI expansion: This move is part of Zoom’s broader strategy to shift from a video conferencing tool to a leader in AI-powered productivity tools for the workplace.

Crowded market dynamics: The AI medical note-taking space has been heating up, with startups like Abridge and Nabla making a name for themselves.

  • Now, larger companies like Zoom are entering the fray, aiming to capture a piece of the growing market. 

  • Competitors like Amazon’s One Medical are also leveraging in-house AI tools to assist clinicians, making this a competitive space.

Why it matters: For healthcare providers, AI-driven tools like Suki save time, allowing more focus on patient care. For Zoom, it’s a chance to become more embedded in the healthcare industry, potentially unlocking new revenue streams and cementing its place in the telehealth ecosystem.

  • Midjourney tests new tools for image editing and re-texturing

  • Stable consistency models reach new AI benchmarks in quality

  • Hugging Face launches SynthID Text for advanced text generation

  • Google DeepMind unveils AI music tools for DJs and creators

  • ElevenLabs releases a feature for creating AI voices from text

  • Solver offers new automated coding API with Elastic Engineering

  • White House under pressure to increase AI use at Pentagon and Intel agencies

Tired of being hands-on-keyboard? Let Claude take control: Join me as I break down how to install and use Claude’s new AI agent.

Ready to make AI work for you? Watch as we show you how to build your first AI business from scratch.

That’s a wrap! Just like Apple is carefully walking the line between reality and AI-enhanced fantasy, we’re all figuring out where to draw the line with new tech. How do you feel about AI’s role in shaping what’s real? Let me know your thoughts!

—Matt (FutureTools.io)

P.S. This newsletter is 100% written by a human. Okay, maybe 96%.