AI pioneer launches latest AGI lab

Good morning, AI enthusiasts. A fresh challenger just entered the AGI arena — and its founder wrote the framework that powers modern AI.

François Chollet’s latest AI lab, Ndea, is approaching artificial general intelligence from a unique approach—but can it outmaneuver the large compute benefits of other tech giants?

Friendly reminder: Our newsletter continues to be slowly moving to a brand new email address. Add [email protected] to your contacts to make sure you keep receiving our emails.

In today’s AI rundown:

• François Chollet founds latest AGI lab

• Microsoft expands Copilot access with free tier

• Try on any clothes virtually with AI

• Luma Labs drops latest next-gen AI video model

• 4 latest AI tools & 4 job opportunities

LATEST DEVELOPMENTS

AI RESEARCH

🚀 François Chollet founds latest AGI lab

The Rundown: François Chollet, former Google researcher and the creator of the favored Keras AI framework, just introduced Ndea, a brand new AI lab aiming to realize AGI through an alternate research method, alongside Zapier founder Mike Knoop.

The small print:

• Ndea’s core strategy combines deep learning with program synthesis, aiming to create AI that may learn and adapt with human-level efficiency.

• The startup positions itself as an alternative choice to the dominant large-scale deep learning approach, arguing that training data limits current AI.

• Ndea plans to construct what they call a “factory for rapid scientific advancement,” specializing in each known frontiers like drug discovery and unexplored territories.

• Chollet also recently launched the ARC Prize Foundation, a nonprofit that’s developing benchmarks to guage human-level AI capabilities.

Why it matters: Chollet is an enormous figure in AI — and his decision to create his own lab could offer a fresh perspective within the race to AGI. With Ndea, Ilya Sutskever’s SSI, and lots of the brightest minds in AI taking different research angles, the groundbreaking achievement could come from any corner of the industry.

TOGETHER WITH HUMAN[X]

🏟️ The long run of AI, multi function place

The Rundown: The HumanX 2025 agenda is officially here, offering a jam-packed lineup of sessions and speakers from firms like OpenAI, Anthropic, and Amazon that dive into AI’s impact on creativity, business, and the human experience.

Attend can’t-miss sessions like:

• How AI will transform the human experience with OpenAI

• The long run of AI starts here: Lessons from Amazon

• Anthropic’s vision for the longer term of AI

• Inside Canva’s dream lab: Designing the longer term of creativity

View the complete agenda and register here using code HX25n_therundown for $150 off a GA pass.

MICROSOFT

🤖 Microsoft expands Copilot access with free tier

The Rundown: Microsoft just introduced Microsoft 365 Copilot Chat, a rebranded free entry-level version of its AI assistant platform that introduces pay-per-use agent capabilities while maintaining core AI features for business users.

The small print:

• The brand new tier offers free access to GPT-4o-powered chat, which incorporates web-based knowledge, file evaluation capabilities, and image and code generation.

• Users can access custom AI agents for task automation, with a consumption-based model at $0.01 per message or $200 for 25,000 messages monthly.

• Agents can leverage knowledge sources for a spread of tasks and actions, and the Copilot Control System allows IT teams to administer the platform easily.

• The offering goals to bridge the gap between free users and the complete Microsoft 365 Copilot subscription ($30/user/month).

Why it matters: Microsoft’s launch is a shot at Gemini’s free push into its Workplace apps, however the differentiator (for now) is the agentic capabilities. For orgs trying to proceed integrating AI easily across their knowledge bases and employees, Microsoft still looks like a powerhouse — though it’ll don’t have any shortage of fast-moving rivals.

AI TRAINING

👔 Try on any clothes virtually

The Rundown: Hugging Face’s latest Leffa space means that you can try on any piece of clothing virtually by uploading your photo and a garment image.

Step-by-step:

1. Visit the Leffa project space on Hugging Face – no registration is required.

2. Upload your full-body photo and the garment you wish to try.

3. In Advanced Options, select your garment type (upper, lower, or dress) and click on Generate to see yourself in latest clothes.

Pro tip: For the very best results, use clear, front-facing photos against easy backgrounds. This is ideal for previewing online shopping purchases.

PRESENTED BY CONTEXTUAL

🤖 Contextual AI releases state-of-the-art RAG platform

The Rundown: Contextual AI, pioneers of Retrieval-Augmented Generation (RAG), just launched the Contextual AI platform to assist organizations construct specialized RAG agents designed for complex, technical, and knowledge-intensive tasks.

With the Contextual AI Platform, you may:

• Construct specialized RAG agents that achieve exceptional accuracy on knowledge-intensive tasks

• Reason over massive volumes of unstructured and structured data

• Maximize user trust with protections against hallucinations and precise citations

Read the complete announcement to learn more and request platform access.

LUMA LABS

🎥 Luma Labs drops latest next-gen AI video model

The Rundown: Luma Labs just released Ray 2, the startup’s next-generation AI video model — which guarantees unprecedented motion quality and physics realism through a brand new multimodal architecture trained with 10x more computing power than its predecessor.

The small print:

• The model can generate high-quality video clips as much as 10 seconds long from text prompts, and it has advanced motion and physics capabilities.

• Ray2 demonstrates a complicated understanding of object interactions, from natural scenes like water physics to complex human movements.

• Ray2 can currently handle text, image, and video-to-video generations, and Luma will soon add editing capabilities to the model.

• The system is launching first in Luma’s Dream Machine platform for paid subscribers, with API access coming soon.

Why it matters: Veo 2’s launch around the vacations felt like a brand new level of realism and quality for AI video, and now Luma punches back with some heat of its own. It’s becoming unattainable to discern AI video from reality — and the query is which lab will crack longer-length, coherent outputs and unlock a brand new realm of creative power.

QUICK HITS

🛠️ Trending AI Tools

• 🧠 Minimax 01 – Ultra-long context open-source LLM assistant

• 🗓️ ChatGPT Tasks – Schedule tasks and reminders in ChatGPT

• 🤖 DryMerge – Automate tasks with AI agents that be just right for you 24/7

• 💪 Master of Pushups – AI vision app to show your phone right into a pushup trainer

💼 AI Job Opportunities

• 📊 Deepmind – Analytics Engineer

• 🛠️ Cresta – Director, Strategic Sourcing and Procurement

• 📈 Databricks – Business Operations Analyst

• 💼 Waymo – Head of FP&A

📰 Every little thing else in AI today

Free event: Learn how to Educate Kids on AI. Hear from a panel of edtech leaders, incl. Sora Schools CEO Garrett Smiley on Jan. 26. RSVP without cost.*

Sakana AI unveiled Transformer², a self-adaptive language model that dynamically adjusts neural pathways for various tasks, outperforming traditional approaches while using fewer parameters.

Google is partnering with the Associated Press to feed real-time news into its Gemini AI assistant, marking the tech giant’s first AI cope with a news publisher.

AI video platform Synthesia secured a $180M Series D, raising the Nvidia-backed company’s valuation to $2.1B.

OpenAI expanded its news partnership with Axios, announcing the funding of 4 latest local newsrooms integrated with ChatGPT and other AI tools.

Cisco announced AI Defense, a brand new safety and security platform that tackles unauthorized AI tampering and data leakage through network-level protection and automatic safety checks.

*Sponsored listing

COMMUNITY

🎥 Join our next live workshop

Join our next workshop this Friday to study how one can prompt ChatGPT o1 models effectively for Coding, Reasoning, and Advanced Research with Dr. Alvaro Cintas, The Rundown’s AI professor.

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

🤝 Share The Rundown, get rewards

We’ll all the time keep this text 100% free. To support our work, consider sharing The Rundown with your folks, and we’ll send you more free goodies.

Your current referral count: 0
What you would like: 1 more referrals until receiving ChatGPT Prompting Kit Bundle

See you soon,

Rowan, Joey, Zach, Alvaro, and Jason—The Rundown’s editorial team