The latest info from Jason and his special guests.
OpenAI Expands Codex: Desktop AI Agent
OpenAI turned Codex into a desktop AI agent that sees your screen, clicks apps, and runs background tasks. Here's what enterprise operators need to know.
Anthropic Releases Opus 4.7: Safer Prelude to Mythos
Claude Opus 4.7 is Anthropic's most powerful generally available AI model — and their safety test bed before Mythos Preview goes public. Here's what operators need to know.
VISOR Turns Visual RAG into an Agentic Workflow
VISOR turns visual RAG into a dynamic agentic workflow with iterative search and over-horizon reasoning. Learn how it solves visual evidence sparsity and search drift — and what enterprise operators need to do now.
GPT-5.4-Cyber: OpenAI vs. Anthropic Mythos
OpenAI launched GPT-5.4-Cyber, a cyber-permissive AI model for security defenders. Is this OpenAI's strategic answer to Anthropic's Claude Mythos Preview? Here's what the AI cyber arms race means for your organization.
Meta Launches Muse Spark: Multi-Agent AI for 3 Billion Users
Meta's Muse Spark launches multi-agent AI to 3 billion users. Learn how parallel subagents, multimodal perception, and the social graph moat change everything for operators.
Claude Mythos Preview Finds Thousands of Zero-Day Exploits
AI Pathfinder | Issue #42 | April 8, 2026 Also published on the AI Pathfinder LinkedIn Newsletter. The cybersecurity landscape just experienced a seismic shift. https://www.youtube.com/watch?v=pT4nHJ48QzQ Anthropic recently dropped two massive announcements: an unreleased frontier model called Claude Mythos Preview, and a massive defensive coalition called Project Glasswing. The reason for the coalition? Mythos Preview is so good at finding and [...]
Top AI Stories This Week: April 6th
Also published on LinkedIn AI Pathfinder Newsletter. Is OpenAI cracking under its own weight? The most anticipated IPO in tech history is turning into a governance disaster in slow motion. If your startup loses three leaders in a week, that is a crisis. If OpenAI does it, that is apparently just Monday. But beneath the surface of the executive shuffle lies [...]
The One-Person Billion Dollar Company Is Here
The holy grail of the AI era has just been achieved. Or has it? For the last two years, Silicon Valley has been whispering about the myth of the “one-person billion-dollar company.” It was a fun thought experiment, but mostly just seen as a theoretical endpoint of AI leverage. But as of this week, it is no longer theoretical. It is [...]
AI Is Leaving Earth: The Race to Build Data Centers in Orbit
AI Pathfinder, Issue #36 For the last two years, the biggest bottleneck in artificial intelligence hasn't been algorithms or data. It's been concrete, steel, and the electrical grid. Training frontier models requires gigawatt-scale power. Data centers face multi-year build delays, grid connection queues, and local zoning battles. The physical infrastructure of Earth is struggling to keep up with the compute demands [...]
Claude Code Source Leak: Anthropic’s Biggest Mistake
AI Pathfinder, Issue #35 The Claude Code Leak: By the Numbers Anthropic just handed the entire AI industry the blueprints to its most advanced agentic system, Claude Code. In what may be the most consequential mistake in agent history, Anthropic accidentally exposed the internal source code for Claude Code, their flagship AI coding assistant. Over 512,000 lines of production-grade TypeScript spread [...]
Claude Code Can Now Use Your Computer. Good or Bad?
AI Pathfinder, Issue #34 The line between AI that writes code and AI that uses it just disappeared. This morning, Anthropic shipped Computer Use for Claude Code. It’s a feature that sounds simple on the surface but represents a fundamental shift in how we interact with AI development tools. Claude can now open your apps, click through your UI, and test [...]
Why Most AI Pilots Die (And What Survivors Do Differently)
Most AI pilots do not die because the models are weak. They die because the operating model around them is weak. And when that happens, the cost is not just a stalled experiment. It is wasted budget, delayed workflow improvement, leadership fatigue, user skepticism, and a harder sell for the next AI initiative that comes behind it. That is the uncomfortable [...]
Blind Coding Is Here: 5 AI Signals You Can’t Ignore
The job of a software developer is about to be completely rewritten. For the last 40 years, the fundamental unit of work has been a human engineer translating a problem into code. That era is over. The new unit of work is the autonomous loop. This week, five distinct signals emerged that show this isn’t a theoretical future. It’s a present-day [...]
Sora Is Dead. And It Has Nothing to Do With Video.
AI Pathfinder, Issue #32 In September 2025, OpenAI launched Sora to the world with a TikTok-style interface, a billion-dollar Disney partnership, and the kind of press coverage that money can't buy. It hit number one in the Apple App Store. Then it flatlined. By late March 2026, OpenAI had quietly pulled Sora from the API, removed it from ChatGPT, and walked [...]
Cisco DefenseClaw: AI Agent Security for Enterprise
OpenClaw exposed 135K+ instances and 800+ malicious skills. Jason Fleagle breaks down Cisco DefenseClaw — the open-source governance layer that finally makes AI agents enterprise-ready.
The Agentic AI Private Equity Playbook
AI/ML deal value in PE tripled from $41.7B to $140.5B in one year. PE-backed firms with systematic AI have nearly 2x ROIC. Jason Fleagle breaks down the 10 high-impact use cases and 4 strategic shifts defining the agentic PE era.
MiniMax M2.7: The AI Model That Builds Itself
MiniMax M2.7 handled 30–50% of its own RL workflow. Jason Fleagle breaks down the self-improving AI model, benchmark results, and your 3-step action plan.
NVIDIA Releases NemoClaw, Open-Source AI Agent Safe for Work
NVIDIA's NemoClaw is the enterprise security layer OpenClaw has been missing. It sandboxes agents, runs models locally, and bridges the cloud securely. Here's why it's the most important announcement from GTC — and your 3-step action plan.
Palantir & NVIDIA Made On-Prem AI the New Standard
Palantir and NVIDIA launched a Sovereign AI OS — a full-stack, on-prem AI data center in a box. 93% of enterprises are already moving AI workloads off the public cloud. Here's what every business leader needs to know.
The AI Infrastructure Race: What Every Business Leader Must Know in 2026
Google spent $32B on Wiz AI. OpenAI and Meta acquired the agent layer. VCs poured $1B into robotics. Here is what the AI infrastructure race means for your business strategy.
The Anthropic AI Safety Report That Made Me Pause — And What It Means for Business
Anthropic released a sobering AI safety report. Here is an honest analysis of what it means for enterprise AI adoption, governance, and responsible deployment.
GPT-5.4: Not a Chatbot — Your New AI Employee
OpenAI GPT-5.4 is a step-change in AI capability. Learn how to deploy it as an autonomous AI employee in your business operations, marketing, and sales workflows.
Harvard Study: AI Can Predict 71% of Active Fund Manager Decisions
A new Harvard study shows AI can predict active fund manager decisions with 71% accuracy. Here is what this means for AI in finance and investment strategy.
Your AI Is Lying to You — How to Make AI More Accurate and Trustworthy
AI hallucinations are costing businesses real money. Learn why your AI tools give confidently wrong answers and the exact techniques to make them dramatically more accurate.
On-Premises AI vs. Cloud AI: The Break-Even Math Has Changed in 2026
The cost of running AI on your own infrastructure has dropped dramatically. Here is the updated break-even analysis every enterprise AI decision-maker needs to see.