The pace of AI and Robotics has been incredible.

So, I share the most important research and developments every week.

Here's everything that happened and how to make sense out of it: Google DeepMind published new work on using reinforcement learning to train robots to be more agile (like soccer players).

They tested a new method of adding disruptive forces and randomness, which led to impressive results compared to the baseline.

https://x.com/GoogleDeepMind/status/1778377999202541642

AI and Robotics have been developing at an incredible pace.

Here are the 15 most important developments that happened this week:

1. NVIDIA revealed Project GR00T, a foundation model to help robots understand the world 2. Jensen Huang also unveiled NVIDIA Blackwell at NVIDIA GTC.

It reduces cost and energy consumption by up to 25x over an H100.

AI and Robotics is moving fast.

So, I share the most important research every week.

Here's everything you need to know and how to make sense out of it: Cognition AI just introduced Devin, an autonomous AI agent that can write entire software projects based on prompts.

In demos, Devin autonomously completed real jobs posted on Upwork by itself.

OpenAI + Figure

conversations with humans, on end-to-end neural networks:

→ OpenAI is providing visual reasoning & language understanding
→ Figure's neural networks are delivering fast, low level, dexterous robot actions

(thread below)
Here's some background:

Two weeks ago, we announced Figure + OpenAI are joining forces to push the boundaries of robot learning

Together we are developing next generation AI models for our humanoid robots

It’s been 13 days & we're excited to share today's update

The world of AI x Robotics is progressing fast.

So, I share the most important research and developments every week.

Here's everything that happened and how to make sense out of it: MIT researchers developed a robot control app for the Apple Vision Pro.

It can stream your head, wrist, and finger movements over wifi to control a robot in real-time.

Setup is as easy as downloading the app and streaming via Python. This is a fun one.

https://x.com/younghyo_park/status/1766274298422161830?s=20

Early in my tech career, I used to make investors sign an NDA before I would even speak to them.

I was worried that if somebody knew my idea, I could be outpaced or out-executed. Secrecy was an asset as it allowed me to build without tipping off any potential competitors.

Over the last decade, my view on this has completely changed. I no longer believe that being secretive was helpful. In fact, it might have been detrimental to hiring and raising capital.

What I have learned is that the only real competitive advantage in technology is the number of product iterations and the progress made between those iterations. Essentially, I value the rate of progress in the highest regard.

I no longer worry about someone copying me, nor do I think it matters. What matters is my team’s ability to execute at the fastest possible rate and to continue innovating over time. I also believe that sharing publicly is net beneficial for many reasons, including hiring top talent who understand your vision, raising capital from investors who are following your story, signing commercial customers, gathering feedback, and the list goes on. So I’ve been building my companies in public and I urge my fellow founders to do the same.

The rate of technological progress is the only true competitive advantage. A lot of investors ask my view on a particular company or sector. I always suggest looking for the company that is progressing the most per unit time. If i was an investor, i would have a heuristic for pace of innovation.

Excited to share: Figure raises $675M at $2.6B Valuation

+ OpenAI & Figure signed a collaboration agreement to develop next generation AI models for robots

Below are the details: @Figure_robot @OpenAI $675M Series B investments from:

- Microsoft
- OpenAI Startup Fund
- NVIDIA
- Jeff Bezos (through Bezos Expeditions)
- Parkway Venture Capital
- Intel Capital
- Align Ventures

AI x Robotics continues to progress incredibly fast.

DeepMind, NVIDIA, Magic, Adobe, Stability AI, OpenAI, Reddit, and Mistral all made major progress this week.

Here's everything you need to know: Google DeepMind published interesting new robotics research.

They fine-tuned an LLM (Palm 2-S) to allow robots to learn faster from human interaction.

The new method achieves the highest improvement in teachability over traditional RAG and LMPC-Skip.

The pace of AI and Robotics has been incredible.

Google, Meta, Berkeley, Carnegie Mellon, OpenAI, and Samsung all made major progress over the past two weeks.

Here's everything you need to know: Google's AI researchers revealed Lumiere, a breakthrough model for AI video.

Lumiere uses a "space-time" neural network to generate clips in a single pass, as opposed to other models that create distant keyframes.

We'll see text-to-movie in 2024.

2024 will be the year of AI and Robotics.

Figure, Google DeepMind, Apple, Alibaba, Microsoft, and OpenAI have all made major progress over the past weeks.

Here's everything you need to know: We achieved a major breakthrough with Figure-01's AI.

It can learn how to self-correct tasks just by watching humans do tasks.

For example, we can now get human data for applications like making coffee, and train the AI system end-to-end.

https://x.com/adcock_brett/status/1743987597301399852?s=20

AI x Robotics is progressing incredibly fast.

Major advancements this week from Berkeley, OpenAI, Channel 1, Mistral, Microsoft, Waymo, Stability, Meta, Stanford, Alibaba, DeepMind, Imperial College, and the biggest AI conference (NeurIPS).

Here's everything you need to know: Berkeley Research released a new paper presenting showing a human locomotion being fully trained in simulation using reinforcement learning.

And as a result, humanoid robots are walking in Berkeley using AI.

AI x Robotics is progressing incredibly fast.

Significant advancements this week from Stanford/USC, Google DeepMind, Meta, Stability AI, Science Robotics, and more.

Here's everything you need to know: A team of researchers from Stanford, USC, and NVIDIA released 'Agent-Driver.'

The team created an agent-based autonomous driving approach that uses LLMs as a central scheduler.

https://x.com/yuewang314/status/1730308633811153105?s=20

I'd pay for a subscription if someone built this 1-person startup idea: The Problem:

→ I build up a backlog of reading materials during the workdays and I'm only able to read them during my off-hours (on a plane, bedtime, etc)

→ Some are too difficult to read on mobile so I like them in physical print. Example of non-mobile, 2-column PDF:

The electric aircraft company I founded in 2018 raised $215M yesterday.

Investors include Boeing, Stellantis, United Airlines.

Posting the investor slide deck: For Context:

5 years ago I founded Archer Aviation

We design and manufacture electric vertical takeoff & landing (eVTOL) aircraft.

These aircraft are 100% electric, take off like a helicopter (vertically), and fly forward like a traditional airplane.

Every week I get asked my cheat codes to raising money.

Since I’ve raised $1B the last 10yrs, people expect I have secrets.

I don’t have secrets.

But I do have a playbook: - You want a repeatable process -

Don’t bank on raising capital via referrals, networking, or events.

It’s impossible to predict & people don’t like sticking their necks out for referrals.

Instead, build a coin operating machine.

The goal: a process that can be repeated

10 years of my life was building software companies.

My last one sold for $100M.

I doubt I’ll ever return to pure-play software.

So here are my cheat codes: Don’t build “the company”

Day 1: Do NOT go out and hire a bunch of people, spend money on logo/brand, purchase ads, or get long-term office space.

Your only goal is to be scrappy getting users, ship product, and find early signs of product market fit.

Figure in-housed all PR.

The last 10 years, I've done the $20k/mo engagement and was never satisfied.

→ In March, Figure emerged from stealth: 4 press articles.

→ Last week, Series A: 8 press articles.

Some notes: The Figure process was led internally by Lee Randaccio - she crushed it:

→ Mapping all relevant journalists (targeted 50)

→ Aligning on Press Release + Assets

→ Interviews / Embargo's / PR Newswire

Today a founder asked me:

→ “What’s your #1 lesson from scaling your last 3 companies?”

My response:

→ “CEO’s need to run product organizations.”

Here’s what that means: Product Organizations are designed from the ground-up to ship useful product.

→ Team headcount is small & lean

→ The team efficiency is high = driving up speed

→ Company has a culture of internal demo’s

The next wave of AI is in your browser.

Chrome extensions will combine...

→ Existing usage of Chrome
→ New usage of AI

Here are 11 high-impact AI extensions: Eesel (@eeselapp)

Automatically organize your Google Docs, Notion pages, and other work documents in your next tab.

Building a startup is insanely hard.

You have to:

→ Come up with ideas
→ Set up an MVP
→ Organize a team
→ Drive your vision
→ Build a great product
→ Sell, sell, sell

It's easy to get overwhelmed.

Here are 6 startup resources to make it easier: Coming Up with Startup Ideas

"How to Come Up With Startup Ideas" by @paulg
paulgraham.com/startupideas.h…

Your opportunity is only as good as your network.

10 tips to build powerful connections: Increase your meeting surface area

Networking is a numbers game.

You won't bat .1000 choosing the right people to help you on your journey.

But if you meet as many people as you can, you'll definitely reach the right ones in time.