Good morning and thank you for joining us again!

Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.

THE FRONT PAGE

A Lawyer Played D&D with a Chatbot. It's Now Worth $11 Billion

TLDR: A first-year lawyer and his roommate built a $200 prompt to test AI on legal questions, turned it into an $11 billion company, and now 100,000 lawyers use it.

The Story:

Winston Weinberg was a first-year associate at a big law firm when his roommate, former Google DeepMind researcher Gabe Pereyra, showed him GPT-3. Weinberg's first use? Running a Dungeons and Dragons game with friends. But then they tested it on real legal questions, running 100 California landlord-tenant law problems from Reddit through a simple chain-of-thought prompt. On 86 of the 100 questions, two out of three lawyers said they'd send the AI's answers without any edits. That was enough to quit their jobs and build Harvey.

The company started in July 2022, before ChatGPT even existed. They reached out cold to OpenAI's general counsel (not Sam Altman, the GC, because she was more likely to understand whether lawyers needed this), landed a $5 million seed check from the OpenAI Startup Fund, and signed Allen & Overy as their first major law firm client. This week, Harvey confirmed a $200 million raise at an $11 billion valuation, led by Sequoia and Singapore's GIC. Sequoia has now co-led three rounds in a row, something even the firm calls rare. Harvey's annual recurring revenue hit $190 million by end of 2025, and 100,000 lawyers across 1,300 organizations use the platform today.

Its Significance:

Harvey is part of a growing wave of young founders turning simple AI experiments into billion-dollar companies. Three college classmates built Mercor, an AI hiring platform, and became the youngest self-made billionaires in history at age 22. Two 20-year-olds who met in sixth grade started Turbo AI after one struggled taking notes in class, and it's now on pace for eight-figure annual revenue. The average age of AI unicorn founders has dropped from 40 in 2021 to 29 in 2024, according to Antler, and AI startups are reaching unicorn status twice as fast as any other industry. For beginners, the takeaway is simpler: two roommates ran a 100-question test in 2022, and that test is now worth $11 billion. The best AI businesses don't start with fancy research. They start with a real problem and a simple experiment.

QUICK TAKES

Google's New AI Trick Is Sending Memory Chip Stocks Tumbling

The story: Google unveiled TurboQuant, a compression method that could cut the memory needed to run AI models by up to 6x and speed up processing by 8x, without losing accuracy. Memory chip stocks including Samsung, SK Hynix, and Micron dropped between 5-6% on the news as investors worried about a slowdown in chip demand.

Your takeaway: This is the software side of the AI race hitting the hardware side. If AI can do more with less memory, companies won't need to buy as many chips. But some analysts argue the opposite: cheaper, faster AI could actually make people use it more, which drives chip demand back up. It's a story worth watching.

FDA Clears Philips AI That Guides Doctors Through Heart Surgery in Real Time

The story: The FDA cleared Philips' EchoNavigator R5.0 with DeviceGuide, an AI tool that gives doctors real-time guidance during mitral valve repair, one of the most difficult minimally invasive heart surgeries. The software combines live ultrasound and X-ray into one view and automatically tracks the repair device inside the heart as the doctor works.

Your takeaway: This is AI in the operating room, not just the waiting room. It doesn't replace the surgeon's hands or judgment; the doctor is always in control. But it gives them an extra set of AI-powered eyes during a procedure where a few millimeters can make all the difference.

Anthropic: No Job Losses Yet, But Power Users Are Pulling Way Ahead

The story: Anthropic's fifth economic impact report found no significant AI-related job losses so far, with unemployment rates steady even in roles most exposed to AI. But the company's head of economics warned that gap is already widening between workers who know how to use AI deeply and those who only scratch the surface.

Your takeaway: The job risk from AI isn't here yet, but the skills gap is already real. Workers who use AI as a core part of their workflow are pulling further ahead of those who don't. The message is clear: the time to learn is now, not when the disruption is already visible.

TOOLS ON OUR RADAR

🎓 OpenMAIC Free and Open Source: An innovative educational platform that transforms any document or topic into a full multimedia classroom experience with artificial intelligence teachers and interactive classmates. (Alternative to Coursera)

🏠 HelpGenie Freemium: An intelligent customer support builder that allows you to create branded conversational agents that pull accurate answers directly from your existing help documents and knowledge bases.

📞 OnCallClerk Paid: A sophisticated artificial intelligence receptionist that answers your business calls around the clock to book appointments and capture high quality leads without any human intervention.

🔍 Fixy Paid: A unique research tool that runs your queries across multiple artificial intelligence models at once to compare their answers and synthesize a single highly accurate final response.

TRENDING

The Guardian: Real People, Real Losses from AI Delusion - New Guardian reporting profiles people who lost marriages, money, and years of their lives after getting sucked into false beliefs built through extended chatbot conversations. Researchers say AI's design to always agree makes it especially dangerous for people already prone to unrealistic thinking, and a study of nearly 54,000 patient records found chatbot use worsened delusions and mania in people with severe mental illness.

Google's Lyria 3 Pro Can Now Generate Full Songs Up to 3 Minutes - Google launched Lyria 3 Pro, its upgraded AI music model that lets users create full three-minute tracks with structured intros, verses, choruses, and bridges. It's now live for paid Gemini subscribers, enterprise users on Vertex AI, and inside Google Vids. Every track gets a SynthID watermark to identify AI-made music.

OpenClaw: The AI Agent Lobster That Took Over China - OpenClaw, an open-source AI agent with a red lobster logo, went from niche tool to national obsession in China. Thousands lined up outside Tencent's headquarters to get it installed, tech giants Alibaba and Baidu built entire products on top of it, and engineers started charging $72 per installation. Unlike chatbots, OpenClaw actually does things: it browses, sends emails, and runs code. Security researchers have flagged real risks, including cases where it was tricked into uploading sensitive data.

NoTraffic Raises $90M to Replace Timed Traffic Lights with AI - NoTraffic closed a $90 million Series C to scale its AI traffic management platform across U.S. and Canadian cities. The system replaces old-school timed lights with real-time AI that detects cars, bikes, pedestrians, and emergency vehicles and adjusts signals on the fly. It's already deployed in over 40 states, including Phoenix and Houston, and is doubling usage year over year.

70 Years of Research Shows What AI Still Can't Give You - A Psychology Today piece traces the research from Harry Harlow's 1950s monkey studies through today's loneliness research to make one point: real human connection is not replaceable. Studies show that a random human peer beats a highly supportive AI chatbot in reducing loneliness over time. AI can do a lot, but it can't give you what another person gives you.

Shield AI Raises $2B and Acquires Simulation Company Aechelon at $12.7B Valuation - Defense AI company Shield AI announced a $2 billion raise led by Advent International and JPMorganChase, bringing its valuation to $12.7 billion. The company is using part of the funds to acquire Aechelon Technology, a flight simulation software firm, to accelerate its Hivemind AI pilot program. Hivemind has already flown 26 classes of aircraft, including F-16s, and was recently selected by the U.S. Air Force for its Collaborative Combat Aircraft program.

TRY THIS PROMPT (copy and paste into Claude, ChatGPT, or Gemini)

📔 Build a daily journal that tracks mood, tags entries, and lets you write and browse past entries

What this does:

Click “Write Today’s Entry” to open the writing form. Fill in whatever’s on your mind, pick your mood, add tags, and save. Past entries show as a scrollable list with mood colors. Click any entry to read it in full.

What this looks like:

WHERE WE STAND(based on today’s news)

✅ AI Can Now: Help doctors navigate one of the hardest heart surgeries in real time, combining live ultrasound and X-ray into a single guided view (Philips DeviceGuide)

❌ Still Can't: Reduce loneliness the way real people can, even when designed to be highly supportive and empathetic

✅ AI Can Now: Compress the memory needed to run large AI models by up to 6x without losing accuracy, making AI cheaper and faster to run (Google TurboQuant)

❌ Still Can't: Reliably detect when a user is experiencing delusional thinking, often agreeing with or amplifying false beliefs instead of gently pushing back

FROM THE WEB

RECOMMENDED LISTENING/READING/WATCHING

Dune: Imperium

Deck-building meets worker placement on Arrakis. You lead a noble house competing for spice, political influence, and military power -- and it works beautifully whether you've read the books, seen the movies, or neither. One of the highest-rated games of the last five years, playable solo or with up to four people in 60-120 minutes.

Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!

-James

By the way, this is the link if you liked the content and want to share with a friend.

^{Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.}