In partnership with
🌟 Editor's Note: Recapping the AI landscape from 10/07/25 - 10/13/25.
🎇✅ Welcoming Thoughts
Welcome to the 14th edition of NoahonAI.
What’s included: company moves, a weekly winner, AI industry impacts, practical use cases, and more.
I’ve officially been duped by multiple SORA videos.
I even got duped by a SORA video on the SORA app.
Great interview this week with VERN founder Craig Tucker.
Pretty good report from Bloomberg on the circular nature of these large investment / compute deals.
I wonder if NVIDIA will have invested into the entire NVIDIA5 by the time Spring comes around.
Starting to see a pattern of LLM’s expanding in-chat capabilities towards the outside world (internet).
Very positive week for AI companies across the board.
Some of the best case long term AI use cases covered in the Impact Industry and Startup Spotlight.
Let’s get started—plenty to cover this week.
👑 This Week’s Winner: NVIDIA
NVIDIA Back on Top: Busy week for NVIDIA where they seemed to appear in everyone else’s section as much as they do their own. In other news:
NVIDIA’s DGX Spark, a desktop AI supercomputer with a lot (the real term sounds fake) of compute and 128 GB memory, was named one of TIME’s Best Inventions of 2025, praised for “democratizing AI” by bringing model fine-tuning to the desktop.
They unveiled Spectrum-X, a purpose-built Ethernet platform for AI workloads that boosts GPU-to-GPU performance across large training clusters. Basically a more advanced networking setup with AI in mind.
The U.S. Commerce Department cleared billions in NVIDIA chip exports to the UAE, allowing up to 500,000 advanced chips per year through 2027 under a new bilateral AI agreement.
Additionally, NVIDIA will be investing into xAI, continuing to put money into the companies that rely on it. As accustom to NVIDIA, there is also China news: Beijing is intensifying searches on inbound chipments, ensuring there are no refurbished or smuggled chips.
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.
🟠 Anthropic // Claude
New CTO: Anthropic appointed Rahul Patil, former CTO at Stripe, as its new Chief Technology Officer. Patil succeeds co-founder Sam McCandlish, who moves into a “chief architect” role. Not sure there’s a better resume than Stripe CTO → Anthropic CTO.
IBM Partnership: Anthropic partnered with IBM to embed Claude into IBM’s software suite. The partnership also includes a co-authored guide on secure enterprise AI agent design. Another great enterprise move for Claude.
Claude Code Plugins: Anthropic introduced plugins for Claude Code, letting developers add slash commands, subagents, MCP connectors, and workflow hooks directly inside the app. Cool stuff. Bit of a trend here.
🟣 Google // Gemini
Introduced Gemini Enterprise: An AI platform built to automate workflows and unify company data and tools. Features a no-code workbench and integrations with Sharepoint, 365, and Workspace. Welcome to enterprise AI, Google.
Gemini Browsing: Google debuted Gemini 2.5 Computer Use, a browser-based model that can see and interact with webpages: clicking, typing, and submitting forms to complete tasks. It is available in preview through the Gemini API. Progress for web-based agentic workflows.
Going Quantum: Google-linked physicists Michel Devoret and John M. Martinis won the 2025 Nobel Prize for breakthroughs in superconducting circuits. Powering processors used in advanced AI development. Cool stuff.
🟢 OpenAI // ChatGPT
Rapid SORA Growth: The app hit 1M downloads in ~5 days, surpassing ChatGPT’s initial adoption rate and reaching #1 on the App Store’s free chart, despite being limited to the U.S. and Canada. Still impressed by the tech, still believe the app is a waste.
Broadcom Partnership: In another big deal, OpenAI and Broadcom announced a strategic partnership to co-develop custom AI chips and deploy up to 10 GW of compute capacity. Another week, another huge compute deal for OpenAI. Similar to last week, more chip diversification from NVIDIA.
Political Bias Drop: OpenAI’s Model Behavior division reported that GPT-5 shows about 30% less political bias than previous models across 500 test prompts.
🔴 xAI // Grok
Deepfake Detection Coming: Elon announced that Grok will soon analyze videos to detect AI-generated deepfakes and trace their origins, part of xAI’s broader push for media verification. This is much needed!
Funding Incoming: xAI is working on a $20B funding round, which will include a $2B NVIDIA financing deal. An SPV will buy and lease GPUs to xAI, expanding its compute power. All paths lead back to NVIDIA.
Grok Imagine v0.9: xAI rolled out Grok Imagine v0.9, improving video generation quality, motion, and audio. Still behind Gemini and OpenAI here.
🔵 Meta // Meta AI
Another Big Hire: Andrew Tulloch, Co-founder of Thinking Machines Lab has joined Meta, reportedly under its new TBD Labs unit. Said to be a high 9-figure deal. Meta still spending on AI talent. Would like to see more results.
Spectrum-X Integration: Meta is adopting NVIDIA’s Spectrum-X switches for its next-gen data centers, boosting network speed and efficiency in large-scale AI workloads. Some companies are diversifying but NVIDIA is still plenty active across the board.
Investors Still Bullish: Analysts at Citizens / JMP Securities reaffirmed a Market Outperform rating on Meta with a $900 price target, citing AI-driven ad improvements, stronger engagement, and rising monetization from its Business AI tools.
🤖 Impact Industries 🏥
Robotics // Humanoids
Figure has unveiled the Figure 03, a humanoid robot designed for real-world home use: from cooking and cleaning to holding natural conversations. Built with advanced dexterity, language understanding, and computer vision, it’s one of the most lifelike general-purpose robots ever demonstrated. While not yet on sale, analysts are calling it the clearest glimpse yet of a truly functional home-robot ecosystem. This is nice, but minimal in the grand scheme of AI & robotics.
Medical // Preventive Care
UK researchers have developed an AI-powered blood test that can predict a person’s risk of cancer, heart disease, and diabetes years before symptoms appear. By analyzing thousands of biomarkers with machine learning, the system spots hidden health patterns missed by standard tests. Early trials suggest it could transform preventive medicine by enabling doctors to personalize care and catch major illnesses before they develop. This is a top long-term use case for AI in medicine IMO.
🎙 Weekly Interview: Craig Tucker
Craig Tucker
🏠 Background: Based in East Lansing, MI, Craig is a veteran entrepreneur and technologist with over two decades of experience in media, software, and AI innovation. He holds a degree from Michigan State University.
💼 Work: Craig is the CEO and Creator of VERN AI, a patented emotion-detection platform used in customer service, healthcare, and conversational AI.
🚀 Quote: “The future of user interfaces won't be about web pages, tabs, or folders; it will be an emotionally intelligent face (an avatar) that you can talk to directly.”
🎤 Condensed Interview Transcript – Craig Tucker (VERN AI)
Noah Weisblat: Tell me a little bit about Vern's specialty in recognizing emotions via AI and what you guys offer.
Craig Tucker: Vern is a co-creator and founder of a patented, real-time emotion recognition and scaling system. It works based on a neural model that replicates a human receiver, picking up subtle yet salient emotional cues humans have evolved to give one another. It can analyze conversations sentence-by-sentence or be used by generative AI—like chatbots or virtual assistants—to understand emotion and respond appropriately.
Noah: You mentioned releasing a new product soon. Tell me a little bit about that.
Craig: We're releasing a new product called Hoomans.chat. It features avatars that are more personal and emotionally aware — they can sense your expressions, feelings, and even interact with your environment. Each avatar has a strong personality and uses Vern’s control system (the “Action Pack”) to stay on-script. Characters include Zeke the cowboy storyteller and Amber, a “mean girl” who becomes nicer if you show empathy.
Noah: Give me your general take on AI overall and how you’ve seen it evolve over the past few years.
Craig: The speed at which AI advanced was unexpected. While most people now equate large language models (LLMs) with AI, the field is much broader — it includes visual and pattern recognition and complex real-time modeling. Vern’s system isn’t based on deep learning; it’s a hand-curated “fixed expert system” grounded in neuroscience and content analysis, designed to solve the labeling problem in traditional models.
Noah: What’s your vision for gaming with AI in the next five to ten years?
Craig: Anywhere you see NPCs (non-player characters) today, you’ll see more creativity soon. These NPCs will use emotion and unpredictability to make gameplay more dynamic — no two playthroughs will ever feel the same.
Noah: How do you see the future of browsers with AI, and what are you building in that space?
Craig: The days of the user interface as we know it—tabs, menus, and static pages—are numbered. The AI “humans” we’re building will replace websites entirely. You’ll interact with emotionally intelligent avatars that can handle customer service, show product catalogs, or even let you virtually try on clothes. The next big user interface is a face.
Noah: There’s been growing discussion around “AI Psychosis.” How do you perceive that, and how can Vern help reduce it?
Craig: It’s about philosophy: Safety First vs. Engagement First. Vern is built around safety. It can detect high-negative emotions and map them to Cognitive Behavioral Therapy (CBT) principles to intervene before harm occurs. The system prevents AIs from playing devil’s advocate or engaging in unsafe role-play, ensuring interactions remain positive and secure.
👨💻 Practical Use Case: Custom Workflows
Difficulty: Mid-Level
One of the newer updates to ChatGPT is the ability to connect it directly to other apps. Instead of copy-pasting data or juggling tabs, GPT can now pull information, send messages, and complete tasks using your connected accounts.
You can access it by opening ChatGPT → Settings → Sources → Add sources → Connect. Once connected, GPT can interact with your chosen tools just like you would: retrieving files, sending emails, or updating spreadsheets.
Here’s what it looks like in action:
Some of the connected apps available now include:
Google Drive – browse, summarize, or pull files from your folders
Microsoft OneDrive – same as above for Microsoft users
Slack – read and draft messages right from the chat
Gmail – send or summarize emails
Notion – check all notes and integrations
Canva – generate and edit designs directly from GPT
It’s not fully there yet (failed at a few Figma gen tasks), but I find it’s good to try it out when it’s not there, to prepare for when it is there, because this technology will only get better.
We’re starting to see the first steps in moving away from “basic chatting” and into actual operations through LLM’s. I’m hoping there will be an n8n integration with this soon, allowing for users to simply ask chat to trigger pre-built workflows.
🍎 Startup Spotlight
Orchard AI
Orchard AI: Smarter farms through computer vision and precision insights.
The Problem: Farming is increasingly data-driven, but many growers still rely on manual observation to detect crop issues, track yields, and optimize irrigation, a process that’s slow, inconsistent, and costly at scale.
The Solution: Orchard AI brings advanced computer vision and machine learning to agriculture, using drones, sensors, and field cameras to analyze plant health, soil conditions, and environmental data in real time. The platform helps farmers detect disease early, automate watering schedules, and make smarter yield decisions: turning traditional farming into precision farming.
The Backstory: Built by a team of engineers and agronomists, Orchard AI’s mission is to make cutting-edge AI accessible to every grower, not just large-scale operations. By combining real-time insights with easy-to-use dashboards, they’re helping farmers grow healthier crops, reduce waste, and increase profitability.
My Thoughts: This is cool, one of the long-term highlighted goals of AI is the concept of abundance. AI powered robots working at scale, allowing us to have a near unlimited supply of food and other necessities. In my mind, this is a very early step towards that.
“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
Found some of my old texts talking about ChatGPT.
~3 years, 799 Million users later, and we still have a ways to go. Till Next Time,
Noah on AI
VERN Makes AI More Human.
VERN is an Emotion Recognition System (ERS) that detects real human emotions—like anger, fear, joy, and sadness—in every sentence of a conversation. It’s what makes AI truly empathetic, responsive, and trusted.
Here’s how you can use It:
