AI Highlights

1. OpenAI o1 signals the rise of AI agents.

We’re witnessing a turning point in AI. OpenAI’s o1 model officially marks the start of “reasoners”—AI that doesn’t just respond but reasons through complex tasks. This is comparable to human-level problem-solving. Sam Altman recently discussed the potential of this model and the trajectory towards fully autonomous AI agents.

Alex’s take: Agents are autonomous systems that can perform actions without human intervention. It took a while to progress from chatbots to reasoners—it’s been almost two years since ChatGPT was released. But Altman expects the progression from reasoners to agents to be MUCH faster. It’s time to accelerate.

2. The future of work with Microsoft Copilot.

Microsoft has launched Wave 2 of 365 Copilot. Here are three major updates that were worth highlighting.

(i) Copilot in Excel: Copilot can now work with both text and numerical data. It also integrates Python, enabling advanced data analysis—like forecasting and machine learning—without coding. I see this as a powerful addition for anyone working with data.

(ii) Copilot in PowerPoint: Use a simple text prompt to create a compelling presentation. You can even reference Word docs to pull in additional context. I wonder how tools like Gamma and Tome will fair who have led the charge so far when it comes to AI-powered presentations?

(iii) Copilot Agents: These look a lot like custom GPTs or Claude projects. You can upload files and tune “Agents” to a specific use case, but I haven’t seen these demonstrate autonomous capabilities yet. I feel Microsoft is riding the “Agent” buzzword with this. The “Actions” element to these “Agents” is coming soon—so I hope I am proven wrong.

3. Hollywood’s AI era begins.

Lionsgate and Runway have joined forces to develop a new AI video model for film production. The model will be trained on Lionsgate’s proprietary catalogue and will assist directors, filmmakers, and creatives in generating and iterating cinematic videos with AI tools. With major franchises like John Wick, Saw, and The Hunger Games under its belt, Lionsgate is positioning itself as a leader in how AI can be integrated into the future of filmmaking.

Alex’s take: This is a bold (yet smart) move, given the amount of pushback from Hollywood on the inclusion of AI in filmmaking. A first-mover advantage partnering with Runway could give Lionsgate the edge over the coming years to shape the direction of AI in Hollywood—I wonder how many other production companies will follow suit?

Content I Enjoyed

Jim Fan on Nvidia’s Embodied AI Lab and Jensen Huang’s Prediction that All Robots will be Autonomous.

AI research is expanding rapidly, and Nvidia is at the forefront of merging artificial intelligence with robotics.

Jim Fan, a senior research scientist at Nvidia, recently was a guest on Sequoia Capital’s “Training Data” podcast.

He discussed Nvidia’s Embodied AI “GEAR” group and their efforts in building humanoid robots capable of performing tasks across both the physical and virtual worlds.

He believes in Jensen Huang’s (CEO of Nvidia) idea that “Everything that moves will eventually be autonomous.”

In ~10 years, there may be as many intelligent robots as iPhones.

I highly recommend you give this one a listen.

Idea I Learned

Turn Your Documents into a Podcast Using NotebookLM.

I’ve been experimenting with Google’s NotebookLM, and honestly, not enough people are talking about how powerful this tool is.

If you haven’t given it a try yet, it’s a game-changer for anyone who regularly works with complex information and needs a faster, smarter way to break it down.

One of my favourite features is “Audio Overview”, which turns your documents into a podcast-style conversation between 2 hosts.

To help you get started, I’ve made a short video tutorial to work you through the process. Check it out before diving in.

Quote to Share

Brett Adcock on humanoid robots and skill acquisition:

“Once one humanoid robot learns a skill, every robot in the fleet will have this acquired.”

Brett Adcock, the Founder and CEO of Figure, was responding to the idea that robots in the future will have specialised skill models they can download and update seamlessly—similar to apps on a smartphone.

This is giving serious Matrix vibes—just like Neo learning Kung Fu.

The possibilities are endless, and it seems like we’re only scratching the surface of what's to come.

Source: Brett Adcock on X

Question to Ponder

Tl;DR: Be cautious about sharing sensitive information.

By default, ChatGPT uses your content to train its models. To prevent this, you can navigate to ChatGPT → click on your profile → settings → data controls → toggle on/off “improve the model for everyone”.

However, if you toggle this off, you won’t be able to save your conversation history.

A neat workaround is to use a “temporary chat”. This chat won’t be used to train OpenAI’s models.

I do not recommend sharing any sensitive data, regardless of these settings. Instead, try using generic examples or anonymise/alter identifying details.

💡 If you enjoyed this issue, share it with a friend.