Good morning, AI enthusiast.

Nvidia has unveiled a new method that provides a massive 53x speed-up for large language models. This breakthrough dramatically reduces the cost of running powerful AI, removing a major barrier to entry for developers.

By making these models significantly cheaper and faster to operate, this technique could democratize access to high-end AI. The key question is: will this trigger a fresh wave of innovation from smaller players who were previously priced out of the market?

In today’s AI recap:

• Nvidia’s 53x AI model speed boost

• Google’s new Gemini image editor

• Anthropic’s ‘Claude for Chrome’ agent

• 8 trending AI Tools

Nvidia's 53x Speed Boost

The Recap: Nvidia released details on Jet-Nemotron, a new technique that accelerates large language models by up to 53 times. This approach dramatically cuts inference costs while matching the accuracy of leading models.

Unpacked:

• The method works by modifying existing pre-trained models, replacing slow full-attention layers with a hyper-efficient design while keeping the model’s core knowledge intact.

• The new architecture delivers up to a 53.6x speedup in generation and a 6.1x boost in prefill time, allowing models to start generating responses much more quickly.

• This efficiency gain can translate to a potential 98% reduction in inference costs, significantly lowering the hardware and capital requirements for new companies entering the AI market.

Bottom line: This development makes powerful AI more accessible to a wider range of developers and businesses. Expect to see a surge of innovation as the cost barriers for building with large models continue to fall.

Big investors are buying this “unlisted” stock

When the founder who sold his last company to Zillow for $120M starts a new venture, people notice. That’s why the same VCs who backed Uber, Venmo, and eBay also invested in Pacaso.

Disrupting the real estate industry once again, Pacaso’s streamlined platform offers co-ownership of premier properties, revamping the $1.3T vacation home market.

And it works. By handing keys to 2,000+ happy homeowners, Pacaso has already made $110M+ in gross profits in their operating history.

Now, after 41% YoY gross profit growth last year alone, they recently reserved the Nasdaq ticker PCSO.

Invest for $2.90/Share

_{Paid advertisement for Pacaso’s Regulation A offering. Read the offering circular at invest.pacaso.com. Reserving a ticker symbol is not a guarantee that the company will go public. Listing on the NASDAQ is subject to approvals.}

AI Tools of the Day

1. 🌐 PERSO.ai - Instantly translate and dub your videos into 30+ languages with hyper-realistic AI voice cloning and perfect lip-sync technology.

2. 🔍 AI Query - Generate complex database queries from plain English, allowing anyone to talk to their data without writing a single line of SQL.

3. 🔌 Eden AI - Integrate dozens of the best AI models for text, image, and audio from any provider through a single, unified API to simplify your development workflow.

4. 📈 ClickFrom.AI - Automatically optimize your e-commerce store to get recommended by large language models like ChatGPT, pioneering the new field of AI Optimization (AIO).

5. 🛠️ Functionize - Revolutionize software testing with an intelligent platform that creates self-healing tests, drastically reducing the maintenance burden on engineering teams.

6. ⚖️ Spellbook - Draft and review complex legal contracts inside Microsoft Word with an AI copilot that spots aggressive terms and suggests missing clauses.

7. 🗺️ Maps GPT - Create custom, interactive maps for any trip or interest by simply describing what you want to discover, from "cozy cafes in Paris" to "all brutalist architecture in Tokyo".

8. 🚀 ShipAny - Launch your AI SaaS startup in hours instead of months using pre-built NextJS templates with integrated authentication, payments, and AI model connections.

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

Google's 'Nano Banana' Unpeeled

The Recap: Google has officially launched its highly-anticipated image model, Gemini 2.5 Flash Image, confirming it as the viral model codenamed 'Nano Banana' that impressed users with its advanced editing and character consistency.

Unpacked:

• The model’s standout feature is its character consistency, which preserves the appearance of a person or pet across multiple complex edits and scenarios.

• It enables advanced creative tasks like multi-image fusion, allowing you to seamlessly blend subjects and styles from different photos into a single new image.

• Before its official release, the model anonymously topped the public LMArena leaderboard for image editing, validating its high performance against other top models.

Bottom line: Google is making complex, multi-step image editing more accessible and intuitive for everyday users. This release directly challenges competitors and pushes the AI industry toward more controllable and reliable creative tools.

AI Training

I Built a Fully-Automated with AI (using n8n + ElevenLabs)

The Recap: In this video, I'm going to show you how to build a podcast fully researched, written, and narrated by AI. We're going to use Firecrawl to scrape local news stories happening in the Austin area, and then we're going to take all those news stories, write a script using AI, and finally use ElevenLabs' new v3 model to generate a final audio file of the local podcast.

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

Anthropic's New Agent

The Recap: Anthropic is testing 'Claude for Chrome,' a new browser extension that empowers its AI to take direct actions on your behalf. The tool lets Claude see your screen, click buttons, and fill forms to automate tasks right from your browser.

Unpacked:

• The extension enables Claude to handle tasks like managing calendars and expense reports, and is now in a limited pilot for 1,000 Max plan users who can join the waitlist.

• Anthropic is transparent about the security challenges, revealing that initial tests showed a 23.6% success rate for prompt injection attacks designed to trick the AI.

• With new safeguards like action confirmations and suspicious pattern detection, Anthropic has reduced the attack success rate to 11.2%, showing progress in making browser agents safer for wider use.

Bottom line: This pilot represents a major step toward mainstream AI agents that act as true digital assistants. Anthropic’s transparent, safety-focused rollout provides a responsible model for developing this powerful new technology.

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).

AI's Entry-Level Problem

The Recap: A new study from Stanford provides the clearest evidence yet that generative AI is disproportionately eliminating jobs for younger, entry-level workers.

Unpacked:

• Since late 2022, payroll data shows that workers aged 22-25 in AI-exposed fields like software and customer service have experienced a 16% relative decline in employment.

• The impact is highly generational, as more experienced workers aged 30 and over in the same fields saw their employment hold steady or even grow.

• The study distinguishes between AI's use for automation, which correlates with job decline, and its use for augmentation, where employment has remained stable.

Bottom line: This data is one of the first major signals showing AI's real-world effect on the labor market. It presents a clear challenge for creating career paths for a generation whose traditional entry points are beginning to disappear.

The Shortlist

Anthropic settled a class-action lawsuit from authors who accused the company of training its Claude models on pirated books, avoiding a trial with potential damages in the billions.

The IETF proposed a new AI-Disclosure HTTP header, creating a standardized, machine-readable way for websites to signal the presence and degree of AI-generated content.

Perplexity launched a revenue-sharing program for publishers, setting aside $42.5 million to compensate partners whose content is used by its Comet web browser and generative AI search results.

A new study found that endoscopists who used AI assistance became 20% worse at spotting abnormalities on their own afterward, raising concerns about skill degradation from overreliance on AI tools.

Signing off,

David, Lucas, Mitchell — The Recap editorial team