Brett Adcock Profile picture
Jul 20 17 tweets 7 min read Read on X
Another massive week of AI and robotics news.

I summarized everything from OpenAI, xAI, Google, Cognition, Figure, AWS, Meta, UBTECH, Pollen Robotics, Agility Robotics, and more.

Here's everything you need to know and how to make sense out of it:
OpenAI dropped ChatGPT Agent, an automated system that can think, plan, and execute complex tasks on a virtual computer

It merges tools like Operator and Deep Research, and can switch between browsing, coding, and doc creation, while connecting to apps
xAI upgraded Grok with anime-inspired 3D AI companions that can hold voice conversations and build relationships with users

Following the launch, the AI assistant became the top-trending app in Japan's app stores

Grok is also set to land on Tesla cars
Google announced a $2.4B tech licensing deal with Windsurf

Windsurf CEO Varun and the team will work on agentic coding for Gemini, powering the assistant to take on GitHub Copilot and other coding agents
Image
A day after Google's deal, Cognition, the company behind coding agent "Devin," announced that it is acquiring the remainder of Windsurf

With this deal, it gets Windsurf's IP, brand, team, $82M in annual revenue, and $100M+ capital
Figure update: We just introduced a 2.3 kWh F.03 battery, engineered and manufactured in-house. Key highlights:

—5 hrs of runtime
—Improved energy density
—2 kW fast charge with active cooling
—Custom BMS to maintain health
—78% cheaper than F.02
AWS launched Kiro, a new agentic IDE to take on Cursor

It combines agentic coding with spec-driven development to bridge the gap between AI prototypes and complex production-ready apps

Free preview now available
Meta announced plans to build superclusters in Louisiana and Ohio to develop AI superintelligence

The first 1GW facility will come online in 2026, while the second, Hyperion, will scale from 2 to 5GW

The co is aiming to invest hundreds of billions of dollars
Image
China's UBTECH unveiled Walker S2, its next-gen humanoid for industrial use cases

The company claims it's the first humanoid that can swap its battery on its own to ensure continuous operation around the clock
Hugging Face's Pollen Robotics team open-sourced a fully 3D-printed robotic hand that costs less than $200

Weighing 400g, "The Amazing Hand" uses a soft TPU shell
and promises 8 DoF with dual hobby servos per finger
Agility Robotics released a clip of its Digit Robot, showing how it can recover from uncomfortable situations, like when someone pulls a rug from under its feet

The co is targeting industrial tasks with its robots, working with giants like Amazon
Hume AI launched EVI-3, a new speech-to-speech model

It doesn't just mimic the voice of the speaker but also their tone, speaking style, nuances, language, and personality.

All it needs is just 15-20s of audio
Ex-Open CTO Mira Murati's Thinking Machines Lab announced $2B in seed funding

Murati said the company is building multimodal AI and will launch its first product in the "next couple of months" with a major open-source component
Image
Lightricks updated its image-to-video LTXV model to support 60-second-long outputs

It streams video live, returning the first second instantly while building scenes

Also includes controls to adjust poses, depth, and style for real-time scene evaluation
Chinese company Leju Robotics released a clip of its new, shorter humanoid, Roban2

The company is targeting light to heavy factory use cases with different form factors

Last, its Kuavo humanoid was seen moving totes at FAW's car factory
We're hiring for BotQ, our new manufacturing division at @Figure_robot:

> Warehouse lead
> Software Test Eng
> End of Line Lead
> Equipment Eng
> Technician
> Manufacturing Eng

Apply here: figure.ai/careers
x.com/adcock_brett/s…
@Figure_robot That's it for this week's AI and Robotics breakdown.

I share the latest research every week, so follow me @adcock_brett for more.

If you found this valuable, consider a like/retweet to spread the word.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Brett Adcock

Brett Adcock Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @adcock_brett

Jul 13
Significant progress in AI and Robotics this week.

So, I summarized everything from xAI, Perplexity, Hugging Face, Figure, OpenAI, Microsoft, Google, Daxo Robotics, UC Berkeley, and more.

Here's everything you need to know and how to make sense out of it:
xAI launched Grok 4 and its multi-agent-powered sibling, Grok 4 Heavy

Both deliver SOTA performance on benchmarks like AIME 2025, Humanity's Last Exam, and Arc-Agi - beating Gemini 2.5 Pro and OpenAI's o3

Also coming via API with 256k context window
Perplexity introduced Comet, its agentic browser for the AI-first web

It embeds the company's answer engine with an always-on assistant that sits in tabs, answers questions, and helps handle tasks like email management

Available on the $200/mo plan
Read 18 tweets
Jul 6
Another huge week of AI and robotics news.

So, I summarized everything from Microsoft, Google, Meta, Perplexity, HeyGen, Chai Discovery, Replit, LimX Dynamics, K-Scale Labs, DJI, Cursor, Sakana AI, and more.

Here's everything you need to know and how to make sense out of it:
Microsoft introduced MAI-DxO, what the company is calling a "step towards medical superintelligence"

The system solved 85.5% of 304 cases vs. just 20% by experienced physicians

It also delivered greater cost savings than human doctors
Google dropped two big updates this week:

—Veo 3 AI video generator is now available globally for Gemini users with the AI Pro plan
—Gemini Live now integrates directly with Google Workspace apps like Calendar and Gmail
Read 19 tweets
Jun 29
Significant progress in AI and Robotics this week.

So, I summarized everything from Google DeepMind, ElevenLabs, Anthropic, Meta, Microsoft, Apptronik, Mistral, Berkley AI, Galbot, and more.

Here's everything you need to know and how to make sense out of it:
Google dropped major AI models this week, starting with Gemini Robotics On-Device

The VLA model runs locally and enables robots to perform a range of two-handed tasks, without the internet

Can also learn new skills with a few dozen demos
Google DeepMind also released:

—AlphGenome: An AI that predicts how gene mutations cause disease
—Gemma 3n: On-device models with multimodal understanding
—Gemini CLI: Open terminal agent
—Magenta RealTime: Live, open music AI
Read 16 tweets
Jun 22
Another huge week of AI and robotics news.

So, I summarized everything from Google, Midjourney, Figure, Cover, OpenAI, 1X, MiniMax, Hexagon, Sunrise Robotics, Gecko Robotics, and more.

Here's everything you need to know and how to make sense out of it:
Google dropped a bunch of AI updates this week:

—New Search Live with voice capability in AI Mode for U.S. users
—Preview of new 2.5 Flash-Lite, fastest and most affordable in the 2.5 family
—General availability of Gemini 2.5 Pro and 2.5 Flash
Midjourney launched their fist AI video generator with V1, an image-to-video system delivering:

—Four 5s clips from an image, which can be extended up to 20s
—Motion/action controls via prompts
—Signature midjourney aesthetic
—25x less cost than rivals
Read 17 tweets
Jun 15
Significant progress in AI and Robotics this week.

So, I summarized everything from OpenAI, Google DeepMind, Mistral, Meta, 1X, Sakana AI, Nvidia, Infinite Machine, Snap, Bytedance, and more.

Here's everything you need to know and how to make sense out of it:
OpenAI dropped several updates this week, including:

—New o3 Pro, thinks longer and beats rivals on PhD-level math & science
—80% price cut for o3
—More natural Advanced Voice, translating for multiple turns
—Projects with deep research, voice mode
Image
Google DeepMind dropped two products:

Extract: A Gemini-powered system, built with the UK govt, to turn planning docs into digital data for urban planning

Weather Lab: An AI-powered platform for early, accurate predictions of storm paths and intensity
Read 19 tweets
Jun 8
Another huge week of AI and robotics news.

So, I summarized everything from OpenAI, Microsoft, Google, Figure, Unitree, BIGAI, UCR Robotics, Agility Robotics, Agibot, and more.

Here's everything you need to know and how to make sense out of it:
OpenAI dropped a series of updates this week, including:

—Custom and ready-to-use ChatGPT data connectors (Google Drive and more)
—Record mode to turn meetings into transcripts with insights
—Internet access for Codex
—Lightweight memory for free users
Microsoft added a free Video Creator on the Bing mobile app, powered by OpenAI's Sora

The company also added native shopping into Copilot with features like review summaries, price tracking, and deal alerts
Read 19 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(