Claude Platform on AWS gives AWS customers a direct way to provision and access Anthropic’s Claude Platform from the AWS Console. The service appears as a partner-managed AWS experience: you start from your AWS account, complete an Anthropic organization setup flow, then return to AWS to create a workspace and generate credentials.
This post is a simple first-run walkthrough. The goal is not to compare every feature or pricing detail, but to show what the onboarding flow looks like end to end, where the handoff between AWS and Anthropic happens, and how to run a quick “hello world” call once access is ready.
Before you start, make sure you are signed in to the AWS account where you want Claude Platform on AWS enabled. If you use AWS Organizations or IAM Identity Center, it is worth being extra deliberate here: the account and role you use during setup determine where the service is activated and where you will manage access later.
Find the New Service in the AWS Console
Start in the AWS Console and search for Claude Platform on AWS from the global search bar. Because this is a new service, searching directly is the fastest way to find the entry point without hunting through service categories.

Open the service result and you will land on the Claude Platform on AWS console page. This is the AWS-managed starting point for the onboarding flow.

Click Get Started to begin the setup process.

The next screen starts the partner signup and onboarding flow. This is where AWS begins the handoff required to create or connect the corresponding Claude Platform organization.

Before continuing, pause for a moment and confirm that you are in the correct AWS account. This is especially important if you regularly switch between personal, sandbox, production, or organization-managed AWS accounts.
After you click Continue, AWS shows a status update while the backend setup runs. This can take a few minutes, so do not worry if the page remains on the progress view briefly.

When the first part of the setup completes, you are redirected into the next step of the Claude Platform onboarding experience. At this point, you are asked to provide the email address for the owner of your organization.
The word “organization” can be a little ambiguous in this flow. In AWS contexts, it often means an AWS Organization. In the Claude Platform context, it appears to refer to the Anthropic-side organization that will own the Claude Platform account and workspace. In many companies these concepts may map cleanly to the same business entity, but they are still separate systems.
In my testing, the form did not accept every email address. The most likely reason is that Anthropic expects an organization-owned domain rather than a public email provider such as Gmail. Another possibility is that the email address may already be associated with a direct Anthropic API account. Either way, use a real work or organization address that should own this setup.

If you leave the flow before completing all of the setup steps, the AWS Console shows an alert and gives you a way to continue where you left off.
If you manage multiple AWS accounts, this is another reason to be careful with the browser session you use. A clean browser window or an incognito/private window can help reduce the chance of mixing AWS sessions, especially if you are switching roles across several accounts.
After you submit a valid email address, the page displays a confirmation message and directs you to check your inbox for the next step.

The email comes from an Anthropic no-reply address and includes a link to Set up your organization. Open that link to continue the Anthropic-side organization setup.

The setup page asks for basic organization and use-case details. The Organization Name field may be prefilled with a value based on your AWS account ID, but you can change it to something more meaningful for your team or environment.

Once you submit the organization details, you should see a confirmation screen indicating that the Anthropic-side setup has completed.

Access
Return to the AWS Console. You should now land on the Access page for Claude Platform on AWS. Before you can sign in to the Claude Platform console using your AWS credentials, you need to create a workspace.

The quick-start path creates a workspace named Quickstart automatically. This workspace becomes the initial place where you can sign in, manage access, and generate credentials for API calls.

From the Access page, choose one of the available roles and click Sign in. This opens the Claude Platform Console and authenticates you through your AWS credentials. If your AWS account uses IAM Identity Center, this will typically be the SSO-assumed role you used to access the AWS Console.
There are a few useful signs that you are in the AWS-managed Claude Platform experience rather than a regular direct Claude Platform login. For example, the footer indicates that the account is managed by AWS. You can also see the workspace name and the AWS Region associated with it.

Workspaces
Back in the AWS Console, the Workspaces page shows the workspace that was automatically created during setup.
At this stage, the AWS-side workspace management options are limited. Most day-to-day Claude Platform configuration happens from the Claude Platform Console itself, using the Access page and Sign in flow described above.

API Keys
The API keys page gives you two credential options:
- Short-lived credentials, valid for 12 hours.
- Long-term API keys, intended for use cases that genuinely need persistent credentials.
Use short-lived credentials where possible, especially for local experiments and quick validation. Create long-term credentials only when the use case requires them, and treat them like any other production secret: store them in a secret manager or protected environment, never hard-code them into source files, and never commit them to a repository.

Hello World
To make a first API call, open Quickstart from the left navigation. This is the Quickstart section in the Claude Platform console, not the workspace name, even though the automatically created workspace is also called Quickstart.

Choose your operating system, SDK, and preferred programming language, then click Generate Code.
The quickstart generator creates short-lived API credentials and shows code snippets you can use immediately. The examples below are lightly adapted for a local Python project using uv and python-dotenv.
CMD Terminal
uv add python-dotenv
uv add "anthropic[aws]"
uv add anthropic
.env file
# Anthropic API credentials
# Claude Platform on AWS - using Anthropic AWS SDK
ANTHROPIC_AWS_API_KEY=aws-external-anthropic-api-key-xxxxxx
ANTHROPIC_AWS_WORKSPACE_ID=wrkspc_01234567890
AWS_REGION=us-east-1
# Claude Platform on AWS - using Anthropic SDK
ANTHROPIC_API_KEY=aws-external-anthropic-api-key-yyyyyyy
ANTHROPIC_BASE_URL=https://aws-external-anthropic.us-east-1.api.aws
ANTHROPIC_WORKSPACE_ID=wrkspc_01234567890
There are two common ways to call the API from Python. The first uses the Anthropic AWS client, which reads the AWS-specific environment variables.
Python file 1 – hello-world-anthropic-aws-sdk.py
from anthropic import AnthropicAWS
from dotenv import load_dotenv
load_dotenv()
client = AnthropicAWS()
message = client.messages.create(
model="claude-sonnet-4-6",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello!"}],
)
print(message)
Output:
python hello-world-anthropic-aws-sdk.py
Message(id='msg_01TqrEWSBstpnxkDDjZkHZLD', container=None, content=[TextBlock(citations=None, text='Hello! How are you doing today? Is there something I can help you with, or did you just want to chat?', type='text')], model='claude-sonnet-4-6', role='assistant', stop_details=None, stop_reason='end_turn', stop_sequence=None, type='message', usage=Usage(cache_creation=CacheCreation(ephemeral_1h_input_tokens=0, ephemeral_5m_input_tokens=0), cache_creation_input_tokens=0, cache_read_input_tokens=0, inference_geo='global', input_tokens=9, output_tokens=31, server_tool_use=None, service_tier='standard'))
The second option uses the standard Anthropic client with the AWS-provided base URL and workspace header.
Python file 2 – hello-world-anthropic-sdk.py
import os
from anthropic import Anthropic
from dotenv import load_dotenv
load_dotenv()
client = Anthropic(
default_headers={"anthropic-workspace-id": os.environ["ANTHROPIC_WORKSPACE_ID"]},
)
message = client.messages.create(
model="claude-sonnet-4-6",
max_tokens=1024,
messages=[{"role": "user", "content": "Hello!"}],
)
print(message)
Output:
python hello-world-anthropic-sdk.py
Message(id='msg_015zgt9kvYmVuBk8vx94ZXWa', container=None, content=[TextBlock(citations=None, text='Hello! How are you doing today? Is there something I can help you with, or did you just want to chat?', type='text')], model='claude-sonnet-4-6', role='assistant', stop_details=None, stop_reason='end_turn', stop_sequence=None, type='message', usage=Usage(cache_creation=CacheCreation(ephemeral_1h_input_tokens=0, ephemeral_5m_input_tokens=0), cache_creation_input_tokens=0, cache_read_input_tokens=0, inference_geo='global', input_tokens=9, output_tokens=31, server_tool_use=None, service_tier='standard'))
At this point, the setup is complete: the Claude Platform on AWS service is enabled in your AWS account, you have a Claude Platform workspace associated with that setup, and you can call Claude using credentials generated through the AWS-managed experience.