Daily Digest: AI engineers get real

Plus: Hardware for Llama 3

March 13, 2024

Want to get in front of 100k AI enthusiasts? Work with us here

Hello folks, here’s what we have today;

PICKS

  1. Cognition Labs, a new AI company, released Devin yesterday. Devin is an AI software engineer, i.e. an agent that works autonomously, just like a human software engineer. BIG DEAL! The demos are crazy and the team behind Cognition is crazier.🍿Our Summary (also below)

  2. Meta just unveiled two huge AI training clusters. The massive GPU clusters Zuck mentioned in earnings this quarter are coming to reality. These are being used to train Llama 3. As usual, there’s also a dose of open-source hardware and software goodness.🍿Our Summary (also below)

  3. ICYMI, Ben’s Bites has a weekly marketing digest. The next digest drops tomorrow. In the meantime, check out last week’s roundup.

from our sponsor

Command Your Workflow: 100 ChatGPT Prompts Tailored to Transform Your Daily Grind

This isn't just another guide; it's your AI-powered secret weapon to navigate the complexities of modern work life. With 100 actionable ways to integrate ChatGPT, you'll turn time-consuming tasks into a breeze and elevate collaboration to new heights.

Ready to work smarter, not harder? Download now for free!

TOP TOOLS

  • HealthKey - Turn 100s of medical records into a simple patient outline.

  • Truffle-1 - A $1299 inference engine designed to run OSS models using just 60 watts. [Hardware]

  • Deepgram Aura - Lightning-fast text-to-speech for voice AI agents.

  • Zep AI - Long-term memory for AI assistants.

  • Tavus for developers - Build human-like replicas and text-to-video with APIs. (+ $18M series A)

  • Twitter Bookmarks - Every last bookmark, categorized with AI.

  • ion design - Convert Figma designs into clean react code.

NEWS

Unclassifieds - short, sponsored links

  • Master AI-Driven Growth: Learn how leaders at Meta, TikTok, Pinterest & OpenAI-backed startups integrate generative AI into their growth marketing strategies. Use BEN10 to get 10% off. Only 60 seats available – Enroll now

  • DaLMatian - text2sql that actually works. Download our free IDE and try to break it!

QUICK BITES

Forget those coding assistants you've been playing with—a whole new breed of AI is entering the game. Cognition Labs just announced Devin, the world's first fully autonomous AI software engineer. Devin doesn’t do suggestions or auto-complete; Devin takes on whole software projects from start to finish.

What is going on here?

Cognition Labs, a new AI company, launches with Devin—an AI agent for software engineering.

What does this mean?

Devin is the next level in AI coding tools, capable of planning complex tasks, learning as it goes, and even fixing its own mistakes. It's like having a tireless coding buddy with its own workspace.

Cognition Labs is evaluating Devin on a new benchmark made of real-world coding issues. Devin clears up a whopping 13.86% of them without any human assistance, compared to 2-5% across the plain GPT-4 and Claude models.

If we go by the demos, Devin’s a builder: Need a website, an app, or even its own AI model fine-tuned? Devin's got you covered. It handles everything from design to deployment and actively collaborates with you. Devin doesn't just help you write code, it diagnoses and fixes bugs autonomously—even in massive, open-source projects.

The team behind Cognition is cracked. They have 10 IOI (International Olympiad in Informatics) gold medals, more than the number of people in the team. They have raised $21 million Series A led by Founders Fund.

Why should I care?

Again, we don’t do fear-mongering (fear is the mind-killer). But there are two things to realize here:

a) Some jobs will be lost, and believing that you’ll be fine if you don’t get used to the new tools is delusional.

b) Software engineering is not dead, but changing. We’re under a shift where AI takes on substantial development work, changing how software teams operate.

For builders, this is the signal that AI agents over these plain models can give you massive improvements in real-world tasks. I’m excited to see Devin for non-programming roles too.

QUICK BITES

Zuck is going big on AI. The massive GPU clusters he mentioned in earning this quarter are coming to reality. They are designed to train the next generation of crazy-smart AI models. Plus, we get a dose of open-source hardware and software goodness.

What is going on here?

Meta just unveiled two huge AI training clusters, sharing details about the design and performance.

What does this mean?

Each of these clusters packs massive compute: 24,000+ GPUs per cluster to be exact. Meta has even bigger ambitions by the end of 2024, aiming for a total of 600k H100 equivalent GPUs. This is fuel for training complex LLMs, like Llama 3 (which is already under training on this new cluster.)

These are built on Meta’s Grand Teton hardware platform (open-sourced, of course) and use PyTorch. Meta is experimenting with network design. One cluster uses RoCE, and the other InfiniBand which will let them figure out the best way to scale up even further in the future.

Why should I care?

Zuck keeps on giving. First the Llama models, and now sharing details about their hardware work too. To be fair, often, a big part of releasing this in public is to attract insane talent. The promise is simple, Meta has the resources you need to do awesome research.

And unlike other labs which are hush-hush about what’s cooking, Meta is open about LLaMA 3 being in the pipeline. No random shocks (looking at you Sora 👀👀).

Ben’s Bites Insights

We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;

  • All 10k+ links we’ve covered, easily filterable (1 referral)

  • 6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)

Reply

or to participate.