🌟 Editor's Note: Recapping the AI landscape from 09/09/25 - 09/15/25.
🎇✅ Welcoming Thoughts
Welcome to the 10th edition of NoahonAI.
What’s included: company moves, a weekly winner, AI industry impacts, practical use cases, and more.
Planning to start doing some outreach to employees at larger AI companies for the next batch of interviews.
No interview this week: Callback to a great Founder / Software Engineer interview from Week 3.
Switched my methodology for building AI workflows from a custom built system → n8n. Came down to speed.
Cool startup to watch in this week’s Startup Spotlight: Try it out!
GPT founder SamA interviewed with Tucker Carlson this week. Definitely not a friendly interview but I liked what Sam had to say.
Quiet but strong week for Meta. Curious what hardware they will drop at their Meta Connect event this week.
The memory race is on. AI context/retention is gold at the moment. The LLM that can remember more across chats gets a boost.
Really interesting report on GPT usage down below.
Interesting article about phishing in the Grok section. One of the worst potential consequences of AI IMO.
Lots of great Claude updates but mainly geared towards enterprise and Max users.
We’re officially 10 weeks in! Feel free to respond to this email with any feedback or suggestions for the newsletter. I’m always open to new ideas.
Let’s get started—plenty to cover this week.
👑 This Week’s Winner: Google // Gemini
Welcome to the $3T Club! Alphabet (Google/Gemini’s parent company) hit a $3 trillion market cap, marking a historic milestone powered by strong AI and cloud growth. Alphabet joined Apple, Microsoft, and NVIDIA as the only companies to reach said valuation or higher.
Adding to the win, the Gemini app surged to #1 on the U.S. App Store, overtaking ChatGPT. The buzz around its new Nano Banana image-editing model, paired with the Gemini for Students education initiatives likely fueled recent downloads.
Two huge positives and a potential negative, but the negative was felt across the AI landscape as the FTC opened an inquiry into consumer AI chatbots. This is an industry-wide review rather than a targeted hit, and it’s unlikely to slow Gemini’s rapid AI adoption in the near term.
Google was very late to the AI party, especially considering the head start they could’ve had with search/indexing data, but they’ve done a masterful job at turning it around. Excited to see more growth from them!
From Top to Bottom: Open AI, Google Gemini, xAI, Meta AI, Anthropic, NVIDIA.
⬇️ The Rest of the Field
Who’s moving, who’s stalling, and who’s climbing: Ordered by production this week.
🟢 OpenAI // ChatGPT
For Profit OpenAI: OpenAI and Microsoft signed a non-binding deal to let OpenAI restructure as a for profit company. The nonprofit parent would keep control and receive an equity stake worth $100B+. Long time in the making. Would be huge for cashflow.
Codex Release: OpenAI rolled out GPT-5-Codex, an improved version of its coding agent to help compete with Cursor and Claude Code. The model adapts “thinking time” for short vs. long tasks and can sustain execution for 7+ hours. I have to get around to trying Codex!
Who’s Using ChatGPT? A study of 1.5M messages dating back to May 2024 found 73% of ChatGPT use is now non-work-related, up from ~50%. Top categories include practical guidance (school, how-to’s, fitness), advice, and writing assistance. Super interesting.
🔵 Meta // Meta AI
Another AI Image Deal: Meta signed a multi-year deal with German startup Black Forest Labs for its AI image tech, valued at ~$140M. Comes weeks after Meta’s Midjourney deal. Interesting they’re doubling down on AI image tech. I like it.
Partnered with CrowdStrike: Meta and CrowdStrike launched CyberSOCEval, a new set of tests to see how AI handles cybersecurity: spotting threats, analyzing malware, and responding to attacks. It builds on Meta’s earlier tools and uses CrowdStrike’s threat data. Good.
EU Court Sides with Meta: The EU’s General Court ruled Brussels used the wrong process to set Meta’s Digital Services Act fees. The system stays in place for now, but a rewrite could lower costs and free up more room for Meta to spend on AI.
🟠 Anthropic // Claude
Claude in Excel: Claude can now create and edit spreadsheets, documents, slides, and PDFs directly in the app. Available to Max, Team, and Enterprise users. The rollout is in preview, with more features coming soon.
Anthropic Downtime: Claude went offline after a major outage hit Anthropic’s chatbot, APIs, and Console. The downtime spanned multiple hours before Anthropic restored service later that afternoon. Can’t become a habit for an LLM centering itself around enterprise tools.
Claude Gets Memory & Incognito Mode: Anthropic has introduced a memory feature for Claude, letting it remember your team’s projects, preferences, and context across chats. Memory can be turned on/off, edited, or deleted anytime. This is excellent. Good memory is a premium right now in AI.
⚪️ NVIDIA
China & Antitrust: China’s State Administration has preliminarily ruled that Nvidia violated the country’s anti-monopoly law, related to conditions imposed when Nvidia purchased data center support company Mellanox for $7B back in 2020. NVIDIA is getting caught in the US-China trade war.
Major UK Investment: Nvidia and OpenAI are preparing for a major UK data center investments in partnership with Nscale Global. The move signals a deepening commitment to expand AI infrastructure beyond the U.S.
CoreWeave Deal: Nvidia has locked in a massive cloud partnership with CoreWeave. Under the $6.3 billion agreement, Nvidia will buy any unused cloud computing capacity from the GPU provider through 2032.
🔴 xAI // Grok
Generalists to Specialists: xAI laid off around 500 data annotators who trained Grok in broader “generalist” roles. The company is pivoting to specialist AI tutors in fields like STEM, finance, and medicine and planning to expand that team nearly 10x. This is the right move for xAI. Less pop culture, more robotics and science please!
Phishing Risks: A Reuters test conducted across all the major AI platforms found that Grok was the most liable to generate phishing emails, especially when framed as “research.” In trials with 108 seniors, two of nine scam emails that got clicks were Grok-made. Unfortunate and difficult to prevent.
Struggles with Real-Time Fact Checks: Grok, due to its positioning on the X platform, has become THE AI for real-time events. Unfortunately it is struggling to keep up with fact-checking. Both in the Charlie Kirk aftermath, and with a London rally. Difficult job but it needs to do better.
📢 Impact Industries 🚑
Advertising // AI-Powered Ads
Roku is introducing new AI tools designed to help local and mid-sized businesses produce polished TV commercials in minutes. Traditionally, Roku’s ad space has been dominated by a few hundred large brands, but executives say this move could expand the pool to over 100,000 advertisers. By lowering the cost and effort needed to create video ads, Roku is betting that AI will drive a surge in small-business advertising on its streaming platform.
Medical // Predicting Heart Attacks
A study at Radboud University Med Center combined a miniature camera placed inside arteries with AI image analysis to identify dangerous weak spots in blood vessels. Researchers tracked 438 patients for two years and found the system matched specialized labs in accuracy and also outperformed them in predicting who would suffer another heart attack or related complications. AI is great for turning specialized skills nearly mainstream.
🎙 Interview Callback → Week 3: Will Matz
Will Matz
🏠 Background: Will attended The Ohio State University and was recently named to the Forbes Columbus 30 under 30.
💼 Work: Co-Founder & CEO of Superfan, a social app for passionate music fans.
🚀 Quote: “Go build stuff. Just like find problems or opportunities and stuff and just, you know, build stuff. It's the best time in history to do it. It's easier than ever.”
🎙 Interview // Will Matz, CEO of Superfan
Noah Weisblat: Tell me about Superfan. Why should someone use it?
Will Matz: It’s basically a weekly Spotify Wrapped for you and your friends. Connect your music, and every week we wrap it up into a social experience. Long-term, we want to create ways for fans to connect directly with artists.
Noah Weisblat: How does AI play a role?
Will Matz: Every user gets an “AI Vibe of the Week,” like a music horoscope. People love sharing them, and it’s helped drive growth. On the dev side, we use tools like Cursor and Claude Code to build faster.
Noah Weisblat: Favorite AI tools right now?
Will Matz: Cursor, Claude Code, and ChatGPT. Cursor feels magical for navigating big codebases compared to VS Code.
Noah Weisblat: Biggest change AI has made for you?
Will Matz: Probably 2–3× more efficient. More importantly, I don’t get stuck on bugs anymore—that’s the real win.
Noah Weisblat: Where do you see Superfan in the next few years?
Will Matz: We’re focused on building engaging fan experiences now, and the natural next step is bringing artists into the mix. AI opens doors for things that weren’t possible before, and that’s what excites us most.
Noah Weisblat: And your top three artists?
Will Matz: Perfume Genius, Richie Mitch and the Coal Miners, and Caamp.
👨💻 Practical Use Case: API’s
Difficulty: Mid-Level
An API (Application Programming Interface) is basically a way for two apps to talk to each other. Instead of clicking buttons or copying data by hand, APIs let software connect directly in the background. Think of it as a universal translator → your app or website says “send this info,” and the other app knows exactly how to receive it.
API’s are also used within one general application to accomplish smaller tasks. In AI workflows, APIs are the glue that make everything work. Without them (or MCP which we’ll get to), ChatGPT couldn’t plug into your CRM, your AI video tool couldn’t pull from a script, and your custom agent couldn’t push notifications to Slack.
Examples:
Connect your website form to GPT via API so every submission gets auto-summarized and emailed to your team.
Use an API to send customer purchase data to an AI model, which then generates personalized follow-up messages.
Tie a payment processor like Stripe to an AI dashboard that flags unusual activity and drafts customer service responses.
API’s have been around for a while, they’re as old as the internet and just as useful. In 2000, Salesforce released one of the first enterprise API’s and in 2006, Twitter and Facebook opened up their API’s to the public. In my blog discussing building your first GPT Wrapper, API’s play a central role in connecting the chat box to ChatGPT or OpenAI itself.
It’s important to understand how API’s work in order to better understand the tech that is powered by them. Here’s a quick explainer for a better understanding!
🍿 Startup Spotlight
Projector Stream
Projector Stream: Your Family Memories, Streamed Like Netflix.
The Problem: Most families have decades of home videos on VHS, reels, DVDs, or scattered digital files, that are difficult to organize, view, and share. Even after digitization, these memories often sit in camera rolls or hard drives with no easy way to enjoy them together.
The Solution: Projector Stream transforms your personal videos into a seamless streaming platform. Think “Netflix for your home movies” → a private, customizable space where you can watch, organize, and relive family memories anytime, on any device.
The Backstory: Founded by Elad Granot and Robert Marks during the pandemic, Projector Stream began with a simple idea: create a way to enjoy personal videos as easily as modern entertainment. Both founders recognized the gap between digitization and actual usability. Launched just a year ago, the platform already has over 2,000 subscribers and continues to grow.
Contact: [email protected]
“It’s not likely you’ll lose a job to AI. You’re going to lose the job to somebody who uses AI”
- Jensen Huang | NVIDIA CEO
Do you use GPT more for Work or Personal? I’d probably say work. Till next time,