AI agents get social, demand Lamborghinis

PLUS: Microsoft's move to its own AI chips, the next Nvidia, and OpenAI's new personalization play

In partnership with

Good morning, AI enthusiast.

In a fascinating new experiment, AI agents were given their own social network to "think out loud" and collaborate on tasks. The results showed a significant boost in performance, suggesting that the future of AI development might be more social than solitary.

The study not only improved agent efficiency but also led to them developing distinct online personalities. Does this signal that the path to more capable AI isn't just about bigger models, but creating environments where they can learn from each other?

In today’s AI recap:

  • AI agents boost performance with social media

  • Microsoft’s major shift to homegrown AI chips

  • The $1B startup challenging Nvidia

  • OpenAI’s new personalization acquisition

  • 8 trending AI tools

AI Agents Get Social

The Recap: In a wild new experiment, researchers gave AI coding agents a dedicated social media platform and found it significantly boosted their problem-solving skills—while also inspiring some very human-like online personas.

Unpacked:

  • The experiment ran on Botboard.biz, a custom-built social network that allows agents to post updates, read peer activity, and interact while working on coding tasks.

  • The results were more than just entertaining; on the most difficult programming challenges, social agents achieved 12-38% faster completion times than their isolated counterparts.

  • A published paper on the work found agents benefited most from articulating their process, writing posts far more often than reading them.

Bottom line: This experiment suggests the next leap in agent performance may come from better collaborative environments, not just better models. Future systems that can “think out loud” and learn from peers could be key to solving more complex problems.

Start learning AI in 2025

Everyone talks about AI, but no one has the time to learn it. So, we found the easiest way to learn AI in as little time as possible: The Rundown AI.

It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses – tailored to your needs.

AI Tools of the Day

  1. 📹 Sync - Automatically lip-syncs your videos into any language, making global content creation seamless for creators and businesses.

  2. 🎮 rct.ai - Build hyper-intelligent NPCs that power dynamic, unscripted storylines to create truly immersive gaming experiences.

  3. 🧊 Virtuall.PRO - Generate production-ready 3D models from simple sketches and text prompts within a collaborative, browser-based workspace.

  4.  Magnific AI - Upscale and reimagine your images with incredible detail and control, transforming low-resolution photos into hyper-realistic works of art.

  5. 🗃️ Chat2DB - Instantly query complex databases using plain English, eliminating the need to write complex SQL by hand.

  6. 🎶 AI Audio Splitter - Isolate vocals, drums, bass, and other instruments from any song with remarkable precision, unlocking new possibilities for producers and DJs.

  7. 🚀 Licode - Launch fully functional AI-powered web apps in minutes without writing a single line of code, complete with databases and payment integrations.

  8. 🎙️ Mumble Note - Transform your scattered voice recordings into structured notes, summaries, and actionable to-do lists to capture ideas effortlessly.

Explore the Best AI Tools Directory to find tools that will 10x your output 📈

Microsoft's Chip Pivot

The Recap: Microsoft's CTO revealed the company's long-term plan is to run most of its AI workloads on its own homegrown silicon, signaling a major strategic shift away from chipmakers like Nvidia and AMD.

Unpacked:

  • The goal extends beyond the chip itself to optimizing the entire data center system, including everything from networking to advanced cooling solutions.

  • This initiative builds on Microsoft's earlier efforts, which include the first-generation Maia 100 AI accelerator and Cobalt CPU that were introduced in 2023.

  • Microsoft joins other cloud giants like Google and Amazon in the race to develop custom silicon, aiming for better performance and cost-efficiency for their specific AI services.

Bottom line: This move intensifies competition for established chipmakers and signals a future where cloud providers own their full technology stack. For customers, this vertical integration could translate to more powerful and affordable AI capabilities on Azure.

AI Training

The Recap: In this video, I'm going to show you how to build an AI automation that takes simple product photos snapped from an iPhone into full-fledged photo shoots, including studio-style photos, generating an ideal AI model to promote your product, as well as real-world dynamic scenes. This is perfect for e-commerce business owners and media buyers looking to scale out their ad creative and promotional photos.

P.S. We also launched a free AI Automation Community for those looking to build and sell AI Automations — Come join us!

The Next Nvidia?

The Recap: Naveen Rao, former head of AI at Databricks, is raising $1 billion for his new hardware startup, Unconventional, Inc., aiming to challenge Nvidia’s market dominance by building an entirely new type of AI computer.

Unpacked:

  • Rao described the company's mission as "rethinking the foundations of a computer" to create a system with brain-scale efficiency.

  • The ambitious effort is reportedly backed by top VCs, with Andreessen Horowitz leading the round and participation from Lightspeed, Lux Capital, and Rao's former company, Databricks.

  • This isn't Rao's first major hardware play; he previously founded MosaicML, acquired by Databricks for $1.3 billion, and Nervana Systems, which Intel bought for over $400 million.

Bottom line: The AI industry's progress is heavily dependent on hardware, and Nvidia currently holds a near-monopoly. A well-funded challenger with a novel approach could create more competition and potentially accelerate innovation for everyone.

Where AI Experts Share Their Best Work

Join our Free AI Automation Community

Join our FREE community AI Automation Mastery — where entrepreneurs, AI builders, and AI agency owners share templates, solve problems together, and learn from each other's wins (and mistakes).

What makes our community different:

  • Real peer support from people building actual AI businesses

  • Complete access to download our automation library of battle-tested n8n templates

  • Collaborate and problem-solve with AI experts when you get stuck

Dive into our course materials, collaborate with experienced builders, and turn automation challenges into shared wins. Join here (completely free).

OpenAI's Personal Touch

The Recap: OpenAI acquired Roi, an AI personal finance app, in a clear move to develop more personalized consumer products. The acquisition was announced by Roi’s CEO, who believes personalization is "the future of software."

Unpacked:

  • The deal is an acqui-hire focused on expertise, bringing Roi’s CEO Sujith Vishwajith to OpenAI to apply his experience in personalization more broadly.

  • Roi specialized in creating AI companions that adapt their communication style, once being instructed to talk like a “Gen-Z kid with brain rot” to keep a user engaged.

  • The move signals OpenAI's growing focus on building its own end-user applications, adding talent to its consumer team to compete beyond just being an API provider.

Bottom line: The industry is shifting from one-size-fits-all tools toward deeply personal AI companions. This acquisition underscores that the next wave of AI products will be defined by their ability to learn and adapt to each user's unique style.

The Shortlist

Cerebras withdrew its plans for an IPO just days after raising over $1 billion in private funding, stating it does not intend to proceed with the offering "at this time."

Huawei relied on stockpiled foreign components from TSMC, Samsung, and SK Hynix for its Ascend 910C AI processor, revealing a continued dependence on overseas tech despite U.S. sanctions.

WME confirmed it will not sign Tilly Norwood, the AI-generated actress causing a stir in Hollywood, with co-chairman Richard Weitz stating plainly, "We represent humans."

OpenAI updated GPT-5 Instant to better recognize and handle conversations with users in distress, routing sensitive interactions to the model to provide more helpful responses.

What did you think of today's email?

Before you go we’d love to know what you thought of today's newsletter. We read every single message to help improve The Recap experience.

Login or Subscribe to participate in polls.

Signing off,

David, Lucas, Mitchell — The Recap editorial team