🧨 Agentic Era Is Wild & It’s Here

I spend a lot of time looking at AI news, and most weeks, it's incremental. 

A new feature here…a small model update there. Even one or two big drops.

But the last week? The landscape just shifted. BIG TIME.

We saw Anthropic officially file for an IPO while warning that AI is close to building itself. 

Google declared the "agentic era" with persistent background agents. 

And we saw OpenAI start turning ChatGPT into a massive superapp while pushing Codex to non-developers faster than engineers.

We're so far past chatbots. 

This is about structural shifts in how work gets done and who owns the infrastructure.

I pulled apart the 10 biggest moves from Anthropic, Google, and OpenAI this week. 

Because if you're building a business right now, you need to know where the tectonic plates are shifting.

How's that for a Los Angeles inspired segue?

Let's get into it...

🏢 Anthropic: Going Public and Raising Alarms

Anthropic made the biggest AI-based financial move of the year while simultaneously dropping a massive warning about the speed of their own tech.

Key Facts

• 📈 The IPO Clock - Anthropic confidentially submitted a draft S-1 to the SEC, signaling a massive upcoming public offering.
• 🤖 Writing Its Own Code - Claude now writes over 80% of Anthropic's own production code, speeding up engineers 8x.
• 🛡️ Infrastructure Defense - Project Glasswing expanded to 150+ orgs to use AI to find critical security flaws in global infrastructure.

The S-1 filing is huge. 

It means Anthropic is confident enough in their revenue and growth to open the books. 

But the timing is wild, because just days later, they published a report that should make everyone consider hitting pause.

They revealed that their "Mythos Preview" model achieved a 52x speedup on a benchmark task compared to a skilled human. 

Engineers are shipping 8x more code per quarter than in 2024. 

And the wildest part? Claude is now authoring 80%+ of the code merged into Anthropic's own codebase.

The paper explicitly warns that recursive self-improvement – aka: AI building its own successors – could arrive sooner than institutions are ready for. 

Sooner than we're all ready for.

They are calling for a coordinated "brake pedal" across AI labs while pushing the gas to the floor. It's exciting and confusing.

Plus, project Glasswing just expanded to cover critical infrastructure like power grids and water systems, finding over 10,000 high-severity security flaws across 150+ partner organizations where those accepted will get early access to their Mythos model which is being looked-at as something beyond comprehension when it comes to security 

Yikes.

The message from Anthropic this week: we are moving fast, we know it's dangerous, and we want to be the ones who protect you from it.

I don't know if this makes me want to get in on the IPO or move to a remote island with no internet. 

🔍 Google: The Background Agents Arrive

Google used I/O 2026 to push AI out of the chat box and into the background of your life.

Key Facts

• 🧠 Gemini 3.5 and Omni - A new frontier model for complex workflows, plus a model that generates video from any input.
• 🕵️ Information Agents - Search now features persistent agents that monitor the web 24/7 and send you updates.
• ⚙️ Antigravity - Search can now build custom dashboards, trackers, and mini-apps on the fly.

Google is officially leaning into the "agentic era." 

They launched Gemini 3.5 Flash, built specifically for complex, multi-step workflows. 

And they dropped Gemini Omni Flash, which can take text, audio, image, or video and generate high-quality video out the other side. 

But the real shift is in Search. 

They are introducing information agents that run in the background 24/7. 

You just tell it what to track, and it sends you updates. 

They also integrated Antigravity, meaning Search doesn't just give you links anymore. It can code a custom dashboard or a mini-app right in the results page, built just for your specific task.

We talked two weeks ago about how search is now fundamentally different, and now with this agentic era, nearly everything else is evolving as well.

And then there's the Gemini Spark announcement, a personal AI agent integrated across Gmail, Docs, and Calendar that anticipates your needs, and Daily Brief, which synthesizes your inbox and calendar into a prioritized morning action plan.

Google's bet is clear: AI should be a layer that runs underneath everything you already do.

My take.

If you're already (and only) using exclusively Google products this is a no-brainer. 

Lean into the shifts so you don't fall behind.

🛠️ OpenAI: The Superapp and The Workforce

OpenAI is fighting back by trying to own the entire professional workflow, from coding to sales.

Key Facts

• 📱 The Superapp Pivot - ChatGPT is getting a massive overhaul to become a superapp with coding tools and partner integrations.
• 💼 Codex for Everyone - 6 new role-specific plugins launched, proving Codex isn't just for engineers anymore.
• 💭 Dreaming V3 - ChatGPT got a major memory upgrade to synthesize your context over years of conversations.

OpenAI is reportedly planning its biggest ChatGPT redesign ever. 

They want to turn it into a superapp to boost revenue ahead of their own potential IPO. With 900M+ weekly active users, they are integrating Codex, AI agents, and partner services like Canva and Booking.com directly into the interface.

It's interesting. 

So many in my circle are migrating to Claude, and while they don't disclose public information like OpenAI, it's believed that ChatGPT still holds the crown on active users.  

But with all the recent Claude improvements, and advancements, we have to ask ourselves.. is it too late for Chat or are they still early?

And speaking of Codex, they just launched 6 role-specific plugins for Sales, Data Analytics, Creative Production, Product Design, Public Equity, and Investment Banking. 

They also launched Codex Sites, a prompt-to-deployed-app builder that will be rolling out to everyone soon. And this feature looks GOOD.

The fascinating data point here? 

Non-developers make up about 20% of Codex users and are growing 3x faster than developers. 

OpenAI realizes the biggest market isn't helping engineers code faster. 

It's helping marketers, analysts, and operators build tools without needing an engineer at all, with models that think like engineers. 

That's going to have a MASSIVE ripple effect.

Finally, they rolled out Dreaming V3, a new memory architecture for ChatGPT.

Instead of just remembering what you explicitly tell it, it synthesizes your entire chat history in the background to keep your context fresh and relevant. 

Oh, and for the Microsoft fans, they launched Scout, their own background agent for Microsoft 365, plus new proprietary models to reduce their reliance on OpenAI. 

The partnership is getting more complicated than a Bravo original series.

– you can exhale now. 

👉 Ask yourself these questions, even if you're not in tech...

1️⃣ Where in your business are you still acting like the "engineer" when you could be the "manager"? If AI can write 80% of Anthropic's code with the SAME tool you have access to, what percentage of your daily operational tasks could it handle right now?

2️⃣ If Google's background agents become the norm, how does your customer find you when they stop searching manually and start letting their agent do the research?

3️⃣ If your team had access to tools that let them build their own dashboards and workflows without waiting on IT, what bottleneck would disappear tomorrow?

🧰 This Is Exactly Why I Built the Creator AI Skills Stack

Here's the thing about all of this agentic news.

None of it works the way it should if your AI doesn't actually know who you are, how you work, or what your business does.

Background agents that write emails for you?
  - Only useful if the AI knows your voice. 

Codex building workflows for your team?
  - Only if it understands your actual workflows. 

ChatGPT remembering your context over time?
  - Not if you haven't given it the right foundation to build from.

That's the gap most of us are sitting in right now. And it's exactly what the Creator AI Skills Stack was built to close.

It's four installable skill files that give your AI the operating system it's been missing:

1. Voice Builder teaches your AI your brand voice in 30 minutes. No more generic output.
2. Workflow Mapper documents how your business actually runs so your AI can operate against real processes, not just principles.
3. Content Operator lets you produce content in your voice every week, batch or single-piece, without starting from scratch.
4. Decision Filter gives you a 5-question framework to evaluate any AI tool, hire, or project before you commit.

I just finished working 1:1 with 20 amazing people in Austin last weekend where I hosted a hackathon with my friend Anna, and the feedback on these was incredible. 

It's a one-time purchase as a downloadable ZIP. 

Install it, run it, keep it forever.

It's discounted at only $47 before it goes up to $97.

Grab the Creator AI Skills Stack 🔗

Four skills, one ZIP, 48-hour money-back guarantee. 

I KNOW you're going to love it.

✌🏻 Wrap Up

The agentic era is here. 

The question is whether your AI is ready, if YOU are ready, and whether you will use this moment as leverage, or watch it pass you by.

P.S. VIP day spots are OPEN for June AND July! I only take 2 max per month and am LOVING working with people at this level. Bonus, with a VIP day I'll even set these files up for you. 

Let's build 🤝