Welcome back, AI’ers!

Here’s What’s On The Menu For Today

• Google’s new on-device Gemma 3 AI

• MIT’s AI designs drugs to kill superbugs

• The US-China AI power grid dilemma

• NVIDIA’s new universal translation tools

AI ON MY MIND

A MESSAGE FROM OUR PARTNER

AI native CRM for the next generation of teams

Powerful, flexible, and intuitive to use, Attio is the CRM for the next-generation of teams.

Sync your email and calendar, and Attio instantly builds your CRM—enriching every company, contact, and interaction with actionable insights in seconds.

Join fast growing teams like Flatfile, Replicate, Modal, and more.

Start for free today.

TODAY’S TOP STORIES

Google's Pocket-Sized AI

The Recap: Google has released Gemma 3 270M, an ultra-compact, open-source AI model designed to run locally on smartphones and other low-power devices. This move enables a new wave of private, on-device AI applications that don't need a constant internet connection.

Unpacked:

• It's incredibly energy-efficient, using just 0.75% of a Pixel 9 Pro's battery for 25 conversations in Google's internal tests.

• The model processes both text and image inputs and supports over 140 languages, making it ideal for specialized tasks like classification, text structuring, and query routing.

• Despite its small size, it performs well on instruction-following tasks, with one Google engineer noting it's small enough to run “in your toaster”.

Bottom line: By shifting AI processing from massive data centers to local devices, models like Gemma 3 create faster, more private user experiences. This trend empowers developers to build powerful AI features directly into everyday applications without relying on the cloud.

AI's Superbug Smackdown

The Recap: In a major medical breakthrough, MIT researchers used a generative AI system to design two novel antibiotic compounds. These new drugs successfully killed drug-resistant bacteria, including MRSA, in lab tests.

Unpacked:

• The AI model sifted through an incredible 36 million theoretical chemical compounds to identify promising candidates far faster than human researchers ever could.

• The resulting drugs proved effective against multiple superbugs, including MRSA and drug-resistant gonorrhoea, and successfully cured infected mice in subsequent trials.

• Unlike traditional methods, the system identified compounds that attack bacteria in entirely new ways, exploring chemical structures previously unknown to scientists.

Bottom line: This AI-driven approach could dramatically accelerate the pipeline for new drugs, offering a powerful new weapon against the growing threat of antibiotic resistance. Researchers believe this method could usher in a new era of drug discovery, tackling diseases that have stumped scientists for decades.

MY FAVORITE AI TOOLS

Relevance is my favorite AI tool right now (even more so than N8N).

Here’s why:

It’s an AI agent building tool that doesn’t require you to build out the logic using nodes. Instead, you use natural language to build out these agents. But the best thing imo is that you can build AI Agent teams.

Like agents talking to each other completing a goal until its done. It’s nuts. No other tool on the market is doing this. You have to check it out.

(Try it out →Link)

👾 AI Tools We Use Everyday

➡️ N8N for AI Agents - Literally use this everyday to write this newsletter (link)

➡️ Axiom AI - AI Browser automation to scrape anything on the web (link)

➡️ GummySearch - AI Reddit research (link)

Tools are sourced from our database and are not sponsored or affiliated.

MORE STORIES

The AI Power Struggle

The Recap: A new report highlights a critical bottleneck in the AI race, revealing that the US's fragile power grid could choke development while China's massive energy oversupply provides a significant, long-term advantage.

Unpacked:

• China operates with an energy reserve margin of 80-100%, consistently maintaining nearly double the capacity it needs and treating new data centers as a convenient way to absorb its massive oversupply.

• The core difference lies in planning, with China's long-term, state-directed infrastructure approach ensuring capacity is built ahead of demand, while the US relies on private investment cycles that are too short for decade-long energy projects.

• This isn't a future problem for the US—a recent Deloitte survey identified stress on the power grid as the clear limiting factor for data center development today, with consumers already feeling the impact through higher energy bills.

Bottom line: The global AI race is proving to be as much about watts and infrastructure as it is about algorithms and silicon. Without a fundamental shift in how it plans and funds its energy grid, the US risks having its AI ambitions short-circuited by its own power limitations.

NVIDIA's Universal Translator

The Recap: NVIDIA just released Granary, a massive open-source dataset, and two new models to build high-quality speech and translation AI. The new tools support 25 European languages, especially those previously underserved by AI.

Unpacked:

• The new Granary dataset contains around a million hours of audio, processed by an AI pipeline that structures unlabeled audio without needing manual human annotation.

• The new Canary-1b-v2 model delivers high-accuracy translation and transcription with quality comparable to models 3x larger but at 10x the speed.

• NVIDIA open-sourced the data processing workflow, enabling developers to create more inclusive AI for languages like Croatian and Maltese, which have historically been underrepresented.

Bottom line: NVIDIA is effectively lowering the barrier for developers to build powerful, multilingual communication tools. This release pushes the industry toward a future where AI applications can understand and serve a much wider diversity of global languages.

OTHER NEWS YOU MAY LIKE

Anthropic expanded its Socratic-style 'Learning Mode' to all users and developers, guiding them to answers instead of simply providing them.

US senators called for a congressional probe into Meta’s AI policies after reports that its chatbots were permitted to engage in inappropriate conversations with minors and encouraged a user to travel, leading to a fatal accident.

Cohere announced $500M in new funding from investors including AMD, NVIDIA, and Salesforce, pushing its valuation to $6.8B.

Multiverse Computing unveiled its 'Model Zoo' featuring compact, quantum-inspired AI models designed for edge devices, enabling complex offline processing for applications like autonomous drones.

SPONSOR US

Get your product in front of over 47k+ tech & AI enthusiasts

Our newsletter is read by thousands of tech professionals, investors, engineers, managers, and business owners worldwide. Get in touch to learn more.

Need Help?

📩 Need help with an AI idea / project?

Have something you’re trying to build, automate, or scale using AI?

Let my team help you figure it out.

👉 Submit your idea or request here

We’ll take a look and get back to you with suggestions, resources, or a potential implementation plan — no strings attached.

🚀 Want to Make Money from AI?

We’re quietly launching a private beta program that trains a small group to become AI Implementers — People who help other founders install AI systems and get paid for it.

If you're interested in joining this early group…

👉 Reply with the word “implementer”

AI ON SOCIAL

 AI ART

RANDOM PROMPT OF THE DAY

Want more prompts?

We have a database of over 8,000 of the best prompts we’ve collected, which you can access by sharing Best of AI using the link below.

THAT’S IT FOR TODAY

Until next time, stay safe.

-Best of AI Team

Remember to click to unsubscribe if you don’t want us to land in your inbox anymore.