In partnership with

Meet America’s Newest $1B Unicorn

A US startup just hit a $1 billion private valuation, joining billion-dollar private companies like SpaceX, OpenAI, and ByteDance. Unlike those other unicorns, you can invest.

Why all the interest? EnergyX’s patented tech can recover up to 3X more lithium than traditional methods. That's a big deal, as demand for lithium is expected to 5X current production levels by 2040. Today, they’re moving toward commercial production, tapping into 100,000+ acres of lithium deposits in Chile, a potential $1.1B annual revenue opportunity at projected market prices.

Right now, you can invest at this pivotal growth stage for $11/share. But only through February 26. Become an early-stage EnergyX shareholder before the deadline.

This is a paid advertisement for EnergyX Regulation A offering. Please read the offering circular at invest.energyx.com. Under Regulation A, a company may change its share price by up to 20% without requalifying the offering with the Securities and Exchange Commission.

Hey,

Last week, we talked about Microsoft losing the PowerPoint war to Claude and the viral panic sweeping the timeline.

But while everyone was distracted by "which model is smarter," the ground shifted.

The war is no longer about benchmarks.
It is about plumbing.

Google quietly rewrote how browsers talk to AI.
Microsoft silently admitted their own models aren't enough for Office.
And Amazon engineers started a revolt against their own tools.

The models that win won't be the smartest.
They will be the ones that are already there.

Here is the signal for the week of Feb 15 to Feb 22, 2026.

1. Google just rewrote the web. Quietly.

Google and Microsoft did something massive between Feb 10 and 13.
And they didn't hold a press conference.

They shipped WebMCP in Chrome 146 Canary.

Think of it as a translator for AI agents.
Before this, agents had to "look" at a website like a human. They used computer vision to find buttons.
That is slow. And expensive.

WebMCP flips the model.
It lets websites expose their functions directly to the browser via navigator.modelContext.

The results are absurd:
67% fewer tokens used.
98% accuracy compared to vision-based agents.

Why did they ship this quietly?
Because it scares publishers.
When Chrome becomes the agent and the protocol, you don't need to visit the website anymore.
Chrome just does the work for you.

Why this matters:
The era of "vision-based" browsing agents is ending before it really started. If your website isn't readable by WebMCP, you will be invisible to the next generation of AI users.

The Move: Download Chrome Canary 146. Open the developer tools. Check if your company's forms are visible to navigator.modelContext. If not, tell your dev team to read the W3C spec now.
(Source: https://developer.chrome.com/blog/webmcp-epp)

2. Microsoft admits defeat (in the fine print).

This is the biggest story nobody is covering.

If you use the new "Agent Mode" in Word, Excel, or PowerPoint, you aren't using OpenAI.
You aren't using a Microsoft model.

Microsoft's own documentation now explicitly states these agents "exclusively use Anthropic's AI models."

Let that sink in.
Microsoft has invested billions in OpenAI. They built their entire brand on Copilot.
But when it came time to ship complex reasoning agents for the enterprise suite?
They outsourced it to Claude.

This rollout is hitting all Microsoft 365 subscribers now. Even the unlicensed ones.
It is a massive architectural signal.
Microsoft cares about utility more than loyalty.

Why this matters:
If Microsoft doesn't trust their own models for complex agentic workflows, neither should you.

The Move: Go to your Microsoft 365 Admin Center. Verify the "Anthropic" provider is enabled. If you block this, you are breaking the best features in Office.
(Source: https://learn.microsoft.com/en-us/copilot/microsoft-365/wordexcelppt-agents)

3. The Amazon Engineer Revolt.

Amazon leadership issued a mandate:
80% of developers must use AI coding tools weekly.

Corporate wanted them to use "Kiro"—Amazon's in-house tool.
The engineers had a different idea.

1,500 Amazon engineers signed a petition endorsing Claude Code instead.
They explicitly flagged Amazon's internal tools as inferior.
This is embarrassing for leadership.
But it is a lesson for every operator reading this.

You cannot mandate a bad tool.
Talent will always find the path of least resistance.
Even if that path costs $8 billion in external investment.

Why this matters:
Standardization is the enemy of velocity. If you force your team to use "approved" tools that suck, they will either quit or shadow-IT their way to success.

The Move: Stop buying enterprise licenses for tools nobody uses. Ask your best developer what they use on the weekend. Buy licenses for that.
(Source: https://www.linkedin.com/posts/amit-maharajh-b2336015_amazon-wants-80-of-its-developers-to-use-activity-7429790797916422144-bRIN)

The Lightning Round

Claude Code Security: Anthropic's new coding agent doesn't just write code; it hunts bugs. It found 500+ zero-day vulnerabilities in open-source projects during testing. It even requires human sign-off before disclosing them.
(Source: https://www.anthropic.com/news/claude-code-security)

Gemini 3.1 Pro: Google's reasoning model is out globally. It is hitting state-of-the-art on GPQA Diamond benchmarks. If you pay for Gemini Advanced, you have it today.
(Source: https://gemini.google/release-notes/)

Jules CI Fixer: Google's coding agent "Jules" now lives in your pull requests. It detects a failed test, fixes the code, and commits the changes in a loop until it passes.
(Source: https://jules.google/docs/changelog/)

Reddit AI Shopping: Reddit is testing AI search that digs up product recommendations and buy links. 80 million people use Reddit to verify purchases—now the AI does the digging for them.
(Source: https://techcrunch.com/2026/02/19/reddit-is-testing-a-new-ai-search-feature-for-shopping/)

What's Happening This Week

If you want to build this stuff with people who are actually doing it, come hang out.

Agent J Skool Community (Free to join):
This Tuesday: The Builder's Hour. We are building a new workflow live.
This Wednesday: Vibe Code Session. Taking an app from idea to live product using Lovable/Bolt.
This Thursday: AI Prompting Masterclass. Learn to write prompts that actually work.

Webinars:
This Thursday (11 AM EST): Lead Magnet Strategy: How to Maximize Growth With TAM, SAM, and ICP Lead Magnets. Free. Register here: https://luma.com/i2tuzzet
This Friday (2 PM EST): Build Your First AI Agent in OpenClaw & n8n. Only 50 spots. Register here: https://luma.com/a1wf8oze

If you want to learn AI automation with a community of builders, join Agent J on Skool — it's free.

The Sunday Action Plan

  1. Check your forms. Install Chrome Canary. If WebMCP can't read your signup form, fix your HTML structure. The agents are coming, and they don't have eyes—they have APIs.

  2. Unblock Anthropic. If you manage Microsoft 365 for your team, ensure the Anthropic connector is live. Do not handicap your team with inferior models just to stay "pure Microsoft."

  3. Audit your code tools. Ask your developers anonymously: "What tool would you use if you weren't forced to use ours?" Then switch to that tool.

The Bottom Line

We spent 2024 and 2025 obsessing over which model was smarter.
2026 is showing us that intelligence is a commodity.

The winner isn't the smartest model.
It's the one that is embedded in your browser, your word processor, and your terminal.

Don't bet on IQ. Bet on distribution.

See you next Sunday,
John

P.S. Reply "HUNTER" and I'll send you the Deep Research Hunter Protocol — the exact process I use to turn AI into a sales research machine.

Reply

Avatar

or to participate

Keep Reading