Intellectually Curious

Claude Managed Agents: From Chat to Cloud-Hosted Teams

Mike Breault

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 4:34

A deep dive into the April 2026 launch of Claude Managed Agents, a move from standalone models to a managed, stateful runtime that handles sandboxing, memory, and multi-agent orchestration. We examine real-world deployments (Rakuten, Asana, Notion), pricing at $0.08 per session hour, and what this means for developers and end users as infrastructure barriers disappear.


Note:  This podcast was AI-generated, and sometimes AI can make mistakes.  Please double-check any critical information.

Sponsored by Embersilk LLC

SPEAKER_01

So uh I spent, gosh, literally my entire weekend trying to assemble this massive IKEA wardrobe.

SPEAKER_00

Oh no, let me guess. Without the instruction manual.

SPEAKER_01

Completely without the manual. I mean, it was just a total nightmare of, you know, tiny screws and random wooden dowels that didn't seem to fit anywhere.

SPEAKER_00

That sounds uh incredibly frustrating.

SPEAKER_01

It really was. But you know, it actually got me thinking: building that complex wardrobe from scratch is exactly how developers used to build AI agents, like just a year ago.

SPEAKER_00

Right. Yeah. It really was that tedious.

SPEAKER_01

Aaron Powell Welcome to Intellectually Curious. Today we are doing a deep dive into the April 2026 launch of Cloud Managed Agents.

SPEAKER_00

Aaron Powell Which honestly turns that frustrating months-long process you just described into, well, a weekend project. Yeah. Or even just a couple of days.

SPEAKER_01

Aaron Powell Exactly. Our mission today is to figure out how AI is shifting from, you know, just being a standalone model you chat with to a fully managed cloud-hosted infrastructure and why that's a massive leap forward for the apps you use every single day.

SPEAKER_00

Aaron Powell It is a huge, huge paradigm shift.

SPEAKER_01

But quickly, before we get into the Wii's, I do want to mention our sponsor, Embersilk.

SPEAKER_00

Yeah, they do great work.

SPEAKER_01

If you need help with AI training or uh automation, integration, software development, Embersilk is absolutely your go-to. They are amazing at uncovering where agents could make the most impact in your business or even your personal life.

SPEAKER_00

Definitely. You can check them out at Embersilk.com for all your AI needs.

SPEAKER_01

Okay, let's unpack this. Because the old way of building agents was basically like building an entire restaurant just to cook a single meal.

SPEAKER_00

Yes. I love that analogy.

SPEAKER_01

I mean, developers were spending months on secure sandboxing and uh error retries and just state management alone.

SPEAKER_00

Right. Because traditionally models have amnesia. Every time you ask them to do a multi-step task, developers had to manually build a memory bank to remind the AI what it was actually doing.

SPEAKER_01

Which takes forever.

SPEAKER_00

Exactly. But what's fascinating here is that managed agents acts as a stateful long-running runtime.

SPEAKER_01

A runtime, meaning uh what exactly?

SPEAKER_00

Meaning Anthropic now autonomously handles all those sandbox execution containers.

SPEAKER_01

Oh, wow.

SPEAKER_00

Yeah, it holds the memory state. And it even coordinates multi-agent workflows. One AI can now realize a task is too big, spin up five subagents to parallelize the work, and the infrastructure just tracks all of it seamlessly.

SPEAKER_01

Aaron Powell So they just give the chef a fully stacked state-of-the-art commercial kitchen. Trevor Burrus, Jr.

SPEAKER_00

Precisely. And because they handle all that underlying infrastructure, we can see real-world innovation just accelerate.

SPEAKER_01

Aaron Powell Well, here's where it gets really interesting. Because I have to ask, does abstracting all this away cost too much? Or, you know, does it limit what teams can actually do in practice?

SPEAKER_00

Aaron Powell You'd think so, right. But no, it's actually highly efficient. You're basically just paying standard token rates plus, get this, only eight cents per session hour for the active runtime.

SPEAKER_01

Aaron Powell Wait, really? Just eight cents an hour?

SPEAKER_00

Yeah. It's incredibly cheap.

SPEAKER_01

Aaron Powell That almost sounds dangerously cheap.

SPEAKER_00

Well, early doctors are already proving how powerful it is. Like Asana is building these AI teammates to collaborate alongside human workers.

SPEAKER_01

Oh, right. I read about that.

SPEAKER_00

And Notion is using it to let users delegate really complex tasks right inside their workspace. But my absolute favorite example comes from Raccoon.

SPEAKER_01

Oh. What are they doing with it?

SPEAKER_00

One of their executives actually noted that this platform makes their power users like Galileo.

SPEAKER_01

Like Galileo, in what sense?

SPEAKER_00

In the sense that it allows a single person to expand their reach across the entire universe of a company.

SPEAKER_01

Oh, that's a cool way to put it.

SPEAKER_00

Right. Because the infrastructure roadblock is gone, Rakuten was able to deploy specialist agents across like finance, engineering, and marketing within a single week.

SPEAKER_01

Wow.

SPEAKER_00

Yeah. A single employee isn't confined to their one specialty anymore. They have a whole team of specialized agents acting as an extension of their own curiosity.

SPEAKER_01

That is just incredible. They bypass the whole infrastructure headache completely and go straight to solving the actual problems. No more building the kitchen from scratch.

SPEAKER_00

They really do. And you know, it leaves you with this deeply optimistic thought to ponder.

SPEAKER_01

I am all ears.

SPEAKER_00

If infrastructure is no longer a barrier, and literally anyone can spin up a team of specialized AI agents in a matter of days, just imagine the incredible world improving solutions humanity will be able to build next.

SPEAKER_01

That is a phenomenal point. The future is looking remarkably bright. Well, if you enjoyed this podcast, please subscribe to the show. Hey, leave us a five star review if you can. It really does help get the word out. Thanks for tuning in.