English

← Back to list

Anthropic launched Managed Agens: The Good Day of Agent Enterprise?

itadol5j

Claude Managed Ages Figure

Anthropic, make a big move.

To be honest, when I saw the news, my first reaction was that the start-up companies that did Agent were probably going to be insomnia.

Anthropic, this guy made another big move yesterday -- a new product called Claude Managed AgesI don't know. It's not a new model, but it's going to help you get rid of Agent's infrastructure.

What do you call this? It's like a copy.

Agent, what's so hard about this?

First of all, there are so many companies known as Agent that have crossed the river. But look at it, most of it's actually a shell on Claude API, and then a few tools to call, even if it's smart.

Agent, what's the hard part?

Not modelling, but infrastructure.I don't know.

You got it:

  • Memory systemOther Organiser
  • Tool organization:Agent needs to know when to use it.
  • Sandbox EnvironmentOther Organiser
  • MultiAgent Coordination: A few Agents how to work together
  • Organisation:Agent can run in the clouds for hours.

These things, each of them, are the nightmares of distributed system engineers. Many of the business teams had a bunch of engineers specializing in this, which resulted in no time for core business.

Anthropic, this wave just blew up the table.

Now Anthropic says:I've got these for you.I don't know.

Claude Managed Ages, that's it. All you need is:

  1. What do you want, Agent?
  2. Or write a YamL profile
  3. Define boundaries (e.g. what tools can be used)
  4. And then... let it run on the clouds of Anthropic.

Infrastructure, memory system, sandbox environment, multiAgent coordination, all packed for you.

The pile of engineers you originally raised to work in infrastructure can now be transferred to core business.

It's not a product, it's a direct filling in the moat of the start-up company.

Check out the Notion.

WIRED sent a presentation, and Notion is already using this as a customer.

Their product manager, Eric Liu, showed how to throw a bunch of assignments to Claude Managed Agent. Agent was working there in silence, and a mission was pushed down for hours.

And then Eric opens Claude Platform's dashboard, and can see each Agent doing what, what tools, how far it goes.

It's not a toy. It works.

What do we do?

Honestly, seeing this news, I think Agent's business team needs to think about it.

You used to sell it:

  • Agent Development Platform
  • Agent Infrastructure Services
  • Agent Transport Tool

Now, Anthropic has packed all this for you, and the price has not been announced, but it won't be too expensive. $30 billionYeah, it was three times last December.

What else can you sell?

I think there may be a few ways:

1. Vertical Area

Anthropic deals with universal infrastructure, but it's business flow, business logic, and it's up to you. In the fields of medicine, law, finance, Agent has to understand business, this Anthropic can't help you for a while.

Agent's 'plugin' ecology

Just like the browser has plugins, Agent needs tools. You can make tools for Agent, like data analysis plugins, document processing plugins, reptile plugs, etc.

3. Agent Counselling and Landing Services

Many companies want to use Agent, but they don't know how to configure, define mission boundaries and design workflows. You can do this "Agent Implementation" service.

4. Monitoring and optimization

Anthropic gave the infrastructure, but Agent ran well and had to be watched. You can do Agent surveillance, log analysis, performance optimization tools.

OpenAI is still busy.

By the way, OpenAI has a platform called Agent. FrontierIt's competing with Anthropic. Both companies are preparing to be listed this year, so they're pushing their products.

This track, it's getting curly.

How do you say start-ups live? Honestly, it's hard. But it's not completely impossible — giants solve generic problems, you solve vertical ones. This pattern is the same in the era of cloud computing, SaaS, AI models.

It's written at the end.

Anthropic pushed Managed Argentinas, and I think there are a few signals:

1. Agent is already a corporate demand.

It's not a toy, it's not a demonstration, it's a work thing. Anthropic earns $30 billion a year, mostly from Claude Platform corporate clients who work with Agent.

2. Infrastructure is being standardized

As in the year AWS standardized servers, now Anthropic standardized the Agent infrastructure. I'll do it later.

3. Agent entrepreneurship team living space is shrinking

The value of the Universal Agent platform is disappearing, and you have to move in the direction of vertical areas, tool ecology, consulting services.

What can I say?

Times have changed. This thing, Agent, goes from "needs a bunch of engineers to build infrastructure," to "write a Yaml that can run."

Are you saying this is good or bad? It's good for the users, for the start-up company... to think about.

Times have changed, either to keep up or to be eliminated.