In the EU
Your prompts and code stay in the EU. Inference runs on Amazon Bedrock in Ireland, with no routing to other regions.
-2.png)
Your teams want Claude.
Make it secure, governed
and in the EU.
Developers and analysts already use Claude on code, documents and customer data. Without a gateway, nobody can say where that data goes, who uses it or what it costs. CloudNation AI Gateway for Claude puts that under your control, without changing how your people work.
Let's talk ambitionsOne sign-in
Your people sign in with Microsoft Entra ID, in the browser. No shared keys, no tokens to copy or renew.
A budget per person
Every user gets a 4-hour and a weekly budget, checked before a request reaches the model.
No seat fee
You pay for Claude as your teams use it. Not for seats that sit idle.
The solution
A managed Claude setup, in our AWS environment or yours.
We deploy an AI gateway on AWS and connect it to Claude models on Amazon Bedrock. Claude Code and Claude Cowork point to that gateway. Your people open the same tools they know, sign in once and get to work.
Behind the scenes, every request is authenticated, checked against that person's budget and logged. You get the control of an enterprise setup with the simplicity of a team subscription. And because you buy through CloudNation, there is no separate contract with Anthropic to set up: one agreement, one invoice.
This is a managed solution: we build it, run it and keep it current. You decide who gets access and how far they can go.
Hosted by CloudNation
The gateway runs in AWS accounts we manage for you. Claude usage, AWS costs and FinOps insight through Vantage come together on one CloudNation invoice. Nothing to set up in your own AWS environment.
In your own AWS account
The gateway runs inside your AWS organization, within your landing zone, security policies and existing AWS agreements. Logs and budgets stay in your account. Still fully managed by CloudNation.
AI usage is scaling. Your controls should lead, not follow.
From pilot to daily work
Claude Code and Cowork are moving from a few enthusiasts to whole teams. Usage and cost grow faster than the controls around them. Put guardrails in place before usage grows, not after.
Residency is on the agenda
Your prompts and code are business data. Customers, auditors and regulators ask where it is processed. Keep inference in the EU and know where every request went.
Pay for use, not for seats
Seat licenses make you size for the maximum before anyone types a prompt. Cost should follow real usage, with a budget per person.
Pull quote:
The question is no longer whether your people use Claude. It is whether you can show where the data goes, who uses it and what it costs.
The questions every organization runs into.
When Claude spreads through an organization, security, IT, finance and procurement all come with their own question. Here is how the AI Gateway answers them.
CISO, Where do our prompts and code actually go?
The problem: Developers use Claude on source code, customer data and contracts. Without a gateway you cannot show where that data is processed, and an audit or customer questionnaire wants a hard answer.
How we solve it: Inference only in the EU, no routing to other regions, and every request logged end to end.
IT, Who is using what?
The problem: Teams work with personal accounts or shared keys. There is no overview for security and audit, and no clean way to end access when someone leaves.
How we solve it: Single sign-on with Microsoft Entra ID, and a portal that shows usage per person and per model. Remove someone in Entra ID and their access ends.
CFO, What will this cost us next month?
The problem: Agentic tools like Claude Code can use a lot of tokens in one session, and one heavy user can use up what a team planned for the month. The cost shows up weeks later.
How we solve it: A budget per person, checked before a request reaches the model, and one cost overview with a forecast in the portal.
Procurement, How do we buy Claude without a contract with Anthropic?
The problem: There is no framework agreement with Anthropic and a credit card is not an acceptable way to buy. Projects stall, or teams buy Claude on their own.
How we solve it: Buy through CloudNation, a supplier you may already know. One agreement and one invoice.
CFO — What do we pay before anyone has typed a prompt?
The problem: Seat-based plans charge a fixed fee per user, whether people use Claude daily or never.
How we solve it: No seat fee. You pay for Claude as it is used, plus the services we deliver around it.
IT, Who keeps this running when new models arrive?
The problem: Models, features and security requirements change fast. Without an owner, the gateway becomes a side project for one engineer.
How we solve it: CloudNation runs it: monitoring, regular reviews of usage, cost, security and access, and a backlog for new models and skills.
Same tools for your people. Full control for you.
As Claude adoption grows, organizations face important questions about data security, access management, operational responsibility and AI spending. CloudNation helps address these challenges through secure architectures, centralized governance and ongoing operational support.
Step 1
Your people
Use Claude Code and Claude Cowork, as they do today.
Step 2
Sign in
Single sign-on with Microsoft Entra ID, in the browser.
Step 3
AI Gateway
Checks the budget, applies the rules, logs every request.
Step 4
Amazon Bedrock
Claude models, served in the EU (Ireland).
All components run on AWS, within the EU.
Built on fully managed AWS services
Amazon Bedrock AgentCore Gateway as the single entry point, Amazon Cognito for sign-in, and serverless services for budgets and logging. No servers, containers or proxies to patch.
Same tools for developers
Claude Code and Claude Cowork point to your gateway. One browser sign-in, with silent token refresh. Your people keep their way of working.
Separate access for automation
Agents and scripts get their own machine access, apart from personal accounts, so automation stays visible and controlled too.
From conversation to live
We map your users, identity setup and residency needs, then deliver a scoped deployment built with infrastructure as code.
The portal
One portal for usage, cost and control.
Every AI Gateway comes with a portal. It is how you see, steer and prove what happens with Claude in your organization. We build it, host it and keep it running.
Who uses what
Every request with who, when, which model, the outcome and the estimated amount. Filter by person, model or outcome and export to CSV.
Budgets per person
A 4-hour and a weekly budget per person, with defaults and personal overrides. See exactly which budget blocked a request.
One cost overview
All Claude spend in one place: over time, per model and per top user, with a forecast against your monthly budget. Cost insight runs on Vantage, our FinOps tool.
Privacy by design
The portal never shows what people asked Claude. Only who, when, which model, the outcome and the amount.
Alerts and reports
A message in Teams or by email when someone reaches 80% of a budget. Export an activity summary for finance and audit.
Skills library
Publish skills and plugins once, with versions and an audit log. Your people install them in the Claude desktop app.
User message
Hit a limit? Your people see why, when their budget resets and what they can do next. No surprises, no support tickets.
Secure and in control, every day.
A gateway is only as good as the team behind it. Our AI Center of Excellence keeps your Claude setup secure, current and under control. It is part of the solution, not an add-on.
Governance
Regular reviews of usage, cost, security and access. We align priorities with you and keep the platform on course.
Planned work
We evolve the platform with you: new models, custom usage windows and reporting, or a shared library of skills and MCP servers.
Unplanned work
A safety net when something needs attention, with a support model we define together with you.
Better every month
Incidents feed back into Governance and Planned Work, so issues are fixed at the root.
Start with the gateway. Grow from there.
The AI Gateway is a solid foundation. When you need more, we build on the same platform:
- A shared library of skills and MCP servers for your teams
- Custom usage windows, limits per team and extra reporting
- Private networking between your environment and the gateway
- Claude inference for your own agents and applications
Building AI agents that run your processes? Explore Agent Lake
Built for teams that are ready to scale Claude.
- Your engineers and analysts use Claude Code or Claude Cowork daily, or want to
- You need EU data residency, an audit trail, or both
- You use Microsoft Entra ID as your identity provider
- You want Claude managed for you, not another platform to build and run yourself
- You are outgrowing a team subscription and want central control without paying per seat
Claude in production
Our Agent Lake framework runs in production at multiple clients, with Claude as the primary model.
Delivered on AWS
AWS Premier Tier Services Partner with the AWS AI Competency.
Anthropic Services Partner
Certified in the Claude Partner Network, with early access to new Claude capabilities.
FAQ
Frequently asked questions about the AI Gateway
1. What is CloudNation AI Gateway for Claude?
A managed Claude setup for your teams. It gives your teams secure, governed access to Claude Code and Claude Cowork, with single sign-on through Microsoft Entra ID, a budget per person, a full audit trail and inference in the EU. It is a managed solution: CloudNation builds it, runs it and keeps it current.
2. Do my developers need to change how they work?
No. Claude Code and Claude Cowork point to the gateway as their model provider, and people sign in once through the browser. The tools stay the same.
3. Where is our data processed?
Claude inference runs on Amazon Bedrock in Ireland, with no cross-region routing. The gateway, logs and budgets all run on AWS in the EU.
4. How do you prevent runaway costs?
Every user gets a 4-hour and a weekly budget, checked before a request reaches the model. Users without a budget are rejected, never unlimited. The portal shows spend per person and per model, with alerts at 80%.
5. Can you see what our people ask Claude?
No. The portal shows who used which model, when, the outcome and the estimated amount. Never the content of prompts or answers.
6. Do we need a contract with Anthropic?
No. You buy through CloudNation, with one agreement and one invoice.
7. Does the gateway run in our own AWS account?
That is your choice. We can host it in AWS accounts we manage, with Claude, AWS costs and FinOps insight on one CloudNation invoice. Or we deploy it in your own AWS account, within your landing zone and policies. Either way, CloudNation manages it.
8. What is the difference between the AI Gateway and Agent Lake?
The AI Gateway gives your people secure, managed access to Claude Code and Claude Cowork. Agent Lake is our framework for building and running AI agents that execute tasks and workflows. Many organizations start with the gateway and add agents later.
9. How do we get started?
Tell us how many people you want to enable. We map your users, identity setup and residency needs and come back with a concrete proposal.
-1.png)
Ready to scale Claude, safely?
We'll map your users, identity setup and residency needs, and show you what it takes to get your teams started.
Chat with us
+31 (0)6 48 016 896
info@cloudnation.nl
