Merry Shipmas 🚢

Plus: AI is running out of fuel

Welcome back! Here’s an interesting new trend: OnlyFans creators are now using AI clones to chat with fans, mimicking the creators’ tone and personality 24/7. It’s efficient, it’s uncanny, and it raises a big question: When your favorite creator isn’t the one behind the screen, is it still the connection you signed up for?

The 12 Days of OpenAI

On the first day of Shipmas, OpenAI gave to me…an AI assistant named Sora. 🎵 The company’s 12 Days of OpenAI event is packed with new tools and updates. Here’s what made the list so far:

  • Sora: a video generation AI that offers expanded tools for creative expression and storytelling (keep reading for my deep dive into top Sora's features)

  • ChatGPT o1: an upgraded chatbot (at a cool $200/month for o1 Pro) promising better performance and a more humanlike experience

  • Canvas: a collaborative AI coding/writing tool that goes beyond simple chat (it’s like brainstorming but supersonic)

What are the early reactions? Some users love the potential of tools like Sora to disrupt media and video. And others were excited about Canvas’s feature expansion to free users once it exits beta.

But: Some say the tools feel overly dependent on OpenAI’s closed ecosystem. This raises broader concerns about accessibility for smaller organizations.

The big picture: On the 12th day of Shipmas, could OpenAI finally deliver AGI? By targeting creativity and team collaboration, OpenAI is staking its claim as a leader in applied AI. And while AGI isn’t under anyone’s tree just yet, events like this show that 2025 might bring some big surprises.

ServiceNow speeds up AI training with Fast-LLM

Via GitHub

Faster AI training just got an enterprise twist. ServiceNow, a company known for streamlining IT and HR operations, is diving into the AI world with Fast-LLM—an open-source framework that cuts training times by 20%.

How it works: Fast-LLM acts as a drop-in replacement for existing AI training systems, meaning businesses can adopt it seamlessly without overhauling their existing infrastructure.

Why it matters: AI training has long been a time sink, requiring enormous computational resources and costs. Fast-LLM slashes training time, lowering the barrier to experimenting with and iterating on AI models.

What’s next: Open-sourcing Fast-LLM is just the beginning. With external contributions expected to enhance its functionality, the framework is poised to evolve rapidly. In the long run, ServiceNow could play a pivotal role in standardizing faster, more accessible AI development practices for all kinds of businesses.

Chatbase Transforms Conversations Into Results

Ready to revolutionize customer interactions? Whether you're automating customer inquiries, driving leads, or scaling your support team's capacity, Chatbase makes it easy to deploy AI agents that truly work for you. Now, your AI agents do more than just chat. They can:

  • Retrieve Live Data: Pull real-time information across tools like Calendly availability, or weather updates.

  • Take Comprehensive Actions: Create Zendesk tickets, or manage customer interactions seamlessly.

  • Integrate Across Platforms: Connect with Calendly, Cal.com, Slack, and beyond. Define custom API endpoints to address your unique business needs, powered by advanced AI models from OpenAI, Anthropic, and Gemini.

Are you ready to take your chatbot up a notch, or several?

AI is running out of data—what happens next?

Via Nature

AI’s biggest challenge might be right around the corner: a data drought. A recent study predicts a 2028 “data crash,” when the availability of training data will no longer keep pace with the demands of large models.

Why it matters: AI systems, especially large language models, rely on vast datasets to train effectively. As we approach the limits of available data, the industry is facing a bottleneck that could stall innovation and increase costs for training future models.

The solution: The answer isn’t just more data—it’s better data. Many leading companies are shifting toward smaller, more specialized models. Instead of relying on sheer volume, these models focus on task-specific performance, paving the way for more efficient AI systems.

The big picture: The looming data crunch puts AI at a crossroads. For years, the narrative has been “bigger is better,” but this crisis is forcing reconsideration. The shift toward precision and efficiency over scale could democratize AI, making it more accessible to industries that don’t have the resources for massive datasets.

  • Google Gemini 2 pushes closer to autonomous AI

  • Google’s Jules coding assistant automates bug fixes

  • Snap’s founder reimagines video chatbots with AI

  • YouTube expands auto-dubbing to knowledge content

  • AI company stirs controversy with rage-bait “stop hiring humans” billboards

  • Character.AI hit with another lawsuit over harmful messages to teens

It’s finally here! Sora, OpenAI’s long-anticipated video generation tool, is now available to the public. Let me walk you step-by-step in how to use it:

Say goodbye to prompt engineering: Leonardo just released its new Flow State feature to help you get from idea → execution fast.

BONUS OpenAI Shipmas Next Wave Episode! Check this one out for the full rundown.

That’s a wrap! See you next week.

—Matt (FutureTools.io)

P.S. This newsletter is 100% written by a human. Okay, maybe 96%.