🤖 AI REVIEW FEBRUARY 21, 2025

^{Read Online | Sign Up | Advertise}

Good morning. Two weeks after breaking up with OpenAI, humanoid maker Figure revealed what they’ve been working on: an AI system that lets robots understand speech and handle objects they’ve never seen before.

Could this technology be the breakthrough that finally brings capable robots into our homes?

 

In today’s AI rundown:

• Figure debuts new system for household robots
• Microsoft’s new AI speeds up protein research
• Create your own AI-powered email assistant
• AI matches decade-long superbug research in days
• 4 new AI tools & 4 job opportunities

LATEST DEVELOPMENTS

FIGURE

 Figure debuts new system for household robots

Image source: Figure

The Rundown: Humanoid robot maker Figure just introduced Helix, a new AI Vision-Language-Action model that lets robots understand voice commands and handle items they’ve never seen before — a major step toward practical household robots.

The details:

• The system combines a 7B-parameter “brain” for understanding and a fast 80M-parameter model for precise movement control.
• Figure demonstrated two robots working together to put away groceries they’d never seen before using natural language commands.
• Helix runs efficiently on basic onboard GPUs and requires just 500 hours of training data, far less than previous approaches.
• The breakthrough comes just weeks after Figure ended its OpenAI partnership, suggesting confidence in their in-house technology.

Why it matters: Robots are already proving capable in industrial settings, but it’s a matter of when, not if humanoid robots will play a significant role in household tasks. Figure’s system and its ability to scale robot learning brings the tech a step closer to being able to reliably handle the mess of unique objects and situations around a home.

TOGETHER WITH TURING

 AI’s next era with DeepMind’s Jeff Dean

The Rundown Turing’s AGI Icons Episode 3 features Jeff Dean, Google DeepMind’s Chief Scientist, and Jonathan Siddarth, CEO of Turing, as they discuss the future of AGI, breakthroughs in Google’s Gemini 2.0, and how AI is transforming software engineering, research, and businesses worldwide.

Watch to learn:

• The evolution of AI from early days to AGI
• Key challenges in scaling intelligence
• The future of AI’s real-world impact

Watch the episode now.

MICROSOFT

 Microsoft’s new AI speeds up protein research

Image source: Microsoft Research

The Rundown: Microsoft Research just released BioEmu-1, a new AI system that can predict how proteins change shape and move — generating thousands of protein structures per hour while matching the accuracy of supercomputer simulations.

The details:

• The system generates protein structure samples 100,000x faster than traditional molecular dynamics, turning months of compute into minutes.
• The model was trained on 200 milliseconds of molecular simulation data, over 9 trillion DNA building blocks, and 750,000 stability measurements.
• Testing showed extreme accuracy in predicting how stable proteins are, matching lab measurements even for proteins it hadn’t seen before.
• Microsoft is making the system freely available to researchers worldwide through Azure AI Foundry Labs.

Why it matters: Is this the week of fast takeoff for AI science? Both Microsoft and Google are dropping model after model that accelerate the scientific research process — turning months or years of work into days (more on that below). Plus, with so many systems being open-sourced to researchers across the globe, it’s likely just the start.

AI TRAINING

 Create your own AI-powered email assistant

The Rundown: In this tutorial, you’ll learn how to build a personal email assistant that can draft and manage emails using n8n’s workflow automation and AI capabilities.

Step-by-step:

1. Visit n8n and create a new workflow.
2. Set up a chat trigger as your workflow’s starting point.
3. Add an AI Agent with OpenAI Chat Model, Window Buffer Memory, and Gmail tool
4. Configure Gmail settings for draft creation with dynamic content using $fromAI(‘placeholder_name’)

Pro tip: We did an extensive workshop showing The Rundown University members how to create your own AI Agent to automate tasks and run local AI models with n8n here.

PRESENTED BY DEEPGRAM

 AI speech-to-text, redefined

The Rundown: Deepgram’s Nova-3 is a new speech-to-text model built to excel in the most challenging enterprise environments, from high-volume call centers to chaotic trading floors — ending your days of wrestling with poor transcription accuracy.

What sets Nova-3 apart:

• Industry-leading accuracy for both live and recorded audio
• The first Voice AI model built for real-time multilingual transcription
• Instant customization to fine-tune vocabulary without extensive retraining

Try Nova-3 today.

GOOGLE

 AI matches decade-long superbug research in days

Image source: Ideogram / The Rundown

The Rundown: Google’s AI co-scientist system just independently reached the same conclusion about bacterial antibiotic resistance as Imperial College researchers — in just 48 hours compared to the team’s decade-long unpublished investigation.

The details:

• The AI identified how bacteria steal virus “tails” to spread resistance genes, matching unpublished findings from a 10-year study.
• The system generated five viable hypotheses, with its top prediction matching the experimental results perfectly.
• Researchers confirmed the AI had no access to their private findings, making the matching conclusion even more significant.
• Google publicly announced the Co-Scientist system yesterday, making it available to researchers through a new testing program.

Why it matters: It didn’t take long for Co-Scientist to already make jaw-dropping news, and it’s just a taste of a future where years of scientific breakthroughs will be compressed into days. This testing also illustrates how AI won’t necessarily replace scientists, but dramatically speed up their discovery and validation process.

QUICK HITS

 Trending AI Tools

•  Proxy 1.0 – AI assistant that can click, scroll, and browse the web
•  R1 1776 – DeepSeek’s R1 reasoning model post-trained by Perplexity AI to remove censorship
•  BuzzClip – Generate viral TikTok AI UGCs in 60 seconds
•  Fleet AI Copilot – AI-driven IT assistant for personalized support and streamlined equipment management

 AI Job Opportunities

•  OpenAI – Network Security Engineer
•  Notable – Account Based Motion Coordinator
•  Writer – Enterprise support agent
•  Glean – Technical Writer

 Everything else in AI today

xAI announced that its new Grok-3 model is now freely available for a limited time, with premium users getting increased usage and early access to advanced features.

COO Brad Lightcap revealed that OpenAI now has 400M weekly active users and 2M paid enterprise customers, with developer usage also doubling over the past 6 months.

NVIDIA partnered with the American Society for Deaf Children to launch ‘Signs’, an AI-powered platform that provides real-time feedback for ASL learners alongside a dataset of 400,000 sign language video clips.

Pika Labs released Pika Swaps, allowing users to easily replace any item or character in a scene with image or text prompts.

Spotify announced the integration of ElevenLab’s AI voice technology into the platform, enabling authors to create and distribute AI-narrated content in 29 languages.

MIT researchers unveiled ‘FragFold’, an AI system that can predict which protein fragments can bind to and inhibit target proteins for drug discovery and cellular biology.

COMMUNITY

 Join our next live workshop

Join our workshop today at 4 PM EST to learn how to use the newest Grok 3 model effectively and how it stands out in the current AI landscape with Dr. Alvaro Cintas, The Rundown’s AI professor.

RSVP here. Not a member yet? Join The Rundown University on a 14-day free trial.

 Share The Rundown, get rewards

We’ll always keep this newsletter 100% free. To support our work, consider sharing The Rundown with your friends, and we’ll send you more free goodies.