Welcome to AI Entrepreneurs

This issue dives into AI’s limitless potential—shattering boundaries, fusing disciplines, and redefining what’s possible. From Spotify’s AI audiobook expansion with ElevenLabs to humanoid robots like Helix and NEO Gamma transforming work and home life, we’re spotlighting the innovations rewriting industries. Plus, explore S* framework’s coding breakthroughs, NVIDIA’s video analytics, and viral video tools like Pikaswaps—your front-row seat to AI’s unstoppable rise.

ELEVENLABS

Spotify Expands AI Audiobooks with ElevenLabs Partnership

Spotify is deepening its AI audiobook offerings through a new collaboration with ElevenLabs, a company known for its AI-generated voices as a cost-effective alternative to traditional studio recordings. This move enables independent authors to produce professional-grade narrations without breaking the bank. While Spotify already supports AI-narrated audiobooks via Findaway Voices, these have been restricted to select partners and required manual review before publishing.

However, the expansion raises key concerns about the future of voice actors and potential copyright issues if AI models are trained on existing recordings. With AI-generated music already making waves on Spotify, this shift in audiobook production signals a broader transformation in audio content creation and distribution.

S*Framework: Boosting AI Code Generation with Smarter Selection

Researchers at UC Berkeley have introduced the S* framework, a novel approach to enhancing AI-powered code generation. By combining parallel scaling (generating multiple solutions at once) with sequential scaling (iteratively refining solutions), S* delivers more reliable and accurate results. It integrates test-time compute and external feedback rather than relying solely on internal reasoning, making it compatible with both LLMs and LRMs.

Key Highlights:

• Combines parallel and sequential scaling for smarter AI-generated code.

• Uses adaptive input synthesis to test and compare solutions.

• Smaller models with S* outperform larger reasoning models without it.

• Designed primarily for programming competition tasks, not full-scale software engineering.

• Improves accuracy, but resource efficiency hasn't been a focus yet.

S* generates test inputs to compare multiple program solutions. AI models test, analyze, and select effective code based on execution outcomes, refining solutions iteratively for stronger outputs. S* could make smaller AI models competitive with larger ones, and its principles might extend to software debugging, automation, and AI-assisted development, enhancing industry efficiency.

FIGURE AI

Helix: The Hive-Mind AI Robot That Learns on the Fly

Figure.ai has d Helix, an AI-powered humanoid robot that can see, understand, and act in real-world environments with minimal training. Using a Vision-Language-Action (VLA) model, Helix enables robots to learn tasks instantly—whether it's putting away groceries or identifying unfamiliar objects. Unlike traditional AI, which requires extensive programming, Helix learns on the go and can control multiple robots under a single AI instance, allowing seamless multi-bot collaboration.

With Figure 02, Helix-powered robots are commercial-ready and designed for real-world tasks—from factory work to home assistance. What makes it revolutionary is its hive-mind learning—once one bot learns a task, all connected bots know it. This marks a major step toward autonomous AI-powered labor, making robots more adaptable, efficient, and capable of working alongside humans in dynamic environments.

1X

NEO Gamma: 1X Technologies Brings Humanoids Closer to Home

1X Technologies has unveiled NEO Gamma, a next-gen humanoid robot designed for home integration, featuring natural human-like walking, improved object handling, and conversational AI. With 10x better hardware reliability and a quieter design, NEO Gamma can squat, sit, and assist with household tasks, making it a step closer to real-world usability. The robot is already being tested in employees' homes, showcasing its potential for cleaning, tutoring, and even medical assistance.

While many see this as a breakthrough in domestic robotics, concerns remain about ethics and the impact on human labor in sectors like nursing and home care. As AI-powered humanoids become more lifelike and capable, the debate over human-robot coexistence is only beginning.

AI HEALTH

AI is Reshaping Healthcare—From Robotic Surgery to Life-Saving Diagnostics : ISSUE # 36

This week, we’re diving into AI’s groundbreaking impact on medicine. From Mount Sinai’s first-in-NYC AI-powered robotic system for prostate treatment to AI-driven breakthroughs in stroke prevention, brain imaging, and enzyme engineering, this issue highlights how AI is no longer just a tool—it’s transforming healthcare as we know it.

Stay ahead with the latest insights shaping the future of medicine—subscribe now!

AI BYTES

Google Veo 2 Drops on Freepik – AI Video Just Got Real

Google just launched Veo 2, its most advanced AI video model, exclusively on Freepik. This powerhouse delivers insanely realistic, smooth, and precise animations, making AI video creation easier than ever. The first 10,000 users get two free video generations, so act fast! While it nails quality, current limits include no reference image support and a 5-second clip cap—but it's a game-changer for AI video accessibility worldwide.

OpenAI’s Operator AI Goes Global – Smarter Automation is Here

OpenAI just rolled out Operator, an AI agent for ChatGPT Pro users, now live in Australia, Brazil, Canada, India, Japan, Singapore, South Korea, and the UK. It automates web-based tasks like booking tickets and making reservations, pushing AI interaction to the next level. Expansion to the EU, Switzerland, Norway, Liechtenstein, and Iceland is in the works—stay tuned!

AI CREATIVITY

Viral Videos Made Easy with Pikaswaps

Pikaswaps from Pika Labs turns videos into something wild. This AI tool swaps objects fast—think a mug becoming a UFO or a dog turning into a knight. Found at Pika.art, it’s simple to use and looks super real. People can make crazy clips with it. Here’s how it works!

Easy Steps to Swap Stuff

• Find a Video: Pick one that’s 5 seconds or longer. Clear light and no wiggles make swaps sharp.

• Point to the Swap: Swipe over the object or type “swap the [thing].” The AI gets it perfect. Add a pic for hard ones—it fixes everything.

• Change It Up: Type a cool idea. Try “swap the chair with a fluffy blue cloud floating in the room.” Done in a snap!

NVIDIA

Upcoming Webinar: Unlocking Video Analytics With AI Agents

This interactive webinar will explore NVIDIA Metropolis, fine-tuning vision language models (VLMs) like Cosmos Nemotron, and customizing AI agents for domain-specific insights. Attendees will gain hands-on knowledge on prompt engineering, AI-powered video analysis, and upcoming GTC 2025 sessions. Join industry experts on Feb 26 & 27, 2025, to elevate your AI-driven decision-making.