The Claude setup that gets you hired in Gen AI


Hello Reader,

When I was a Distinguished Cloud Architect at Verizon, I wanted to deploy a new AWS service for our applications. The service was awesome, solved a real problem, and the app team loved it. Security said no. Not because the service was bad, but because data would leave our controlled environment.

We ended up building a workaround that took three extra months. This is the reality at every bank, telecom, healthcare company, and government agency. Cool tech is not enough. Security controls are the gatekeeper.

I am telling you this story because Anthropic and AWS just announced something that finally breaks this pattern for GenAI coding assistants. Claude Code and Claude Co-Work can now run natively on AWS with all the features, pointed at Claude models hosted in Amazon Bedrock. If you work at a regulated enterprise, this is the announcement you have been waiting for.

In today's edition, we will go over what changed, why it matters, how to set it up, and most importantly, how you as a Solutions Architect can take advantage of this in your career.

The Old Way, and Why It Did Not Work for Enterprises

Here is how Claude Code/Cowork has been working until now:

  • You run Claude Code on your laptop
  • You select a model like Opus 4.5 or 4.7
  • The model itself runs on Anthropic's hardware
  • Your prompts, your context, code, your company data, all of it travels to Anthropic's servers

For solo developers and startups, this flow is fine. For enterprises, it is a non-starter. Let me explain why:

  • Security blocker: This is the big one! Your code, prompts, and potentially sensitive data are leaving your infrastructure. If you work in banking, defense, telecom, or healthcare, your security team will reject this in the first review.
  • Cost management: You are paying AWS for your cloud, and paying Anthropic separately for model usage. Two bills, two contracts, two cost centers.
  • No enterprise controls: No IAM, no SSO, no centralized access management, no usage auditing.

Meanwhile, AWS already had Claude models running inside Amazon Bedrock. Opus, Sonnet, Haiku, all there, all secure. But Claude Code could not talk to them. The two worlds were disconnected.

What Changed

With the new announcement, you can point your local Claude Platform directly at Claude models running in your own Amazon Bedrock service. Nothing goes to Anthropic's hardware. The Claude Platform harness runs on your laptop, the model runs on your AWS account, and the loop stays inside your enterprise perimeter.

Why This Is a Big Deal for Enterprises and SAs

Let's look at what you gain:

  • Data stays in your infrastructure. Your prompts and code context hit Bedrock in your account, not Anthropic.
  • Disable metadata collection. By default Anthropic collects usage statistics like token counts. You can turn this off for full security compliance.
  • Single cost surface. Bedrock usage shows up in AWS Cost Explorer. If your company has enterprise discount programs or budget controls, they all apply now.
  • Enterprise access controls. Use AWS SSO, IAM roles, or bearer tokens to control who can invoke which model. Same governance model your company already uses for every other AWS service.
  • Keep the Claude Code harness. Let's be honest, Claude Code and Claude Co-Work have the best agent harness in the market right now. The orchestration, the sub-agent spawning, the tool use, all of it stays. You are not sacrificing features.
  • MCP servers still work. You can still connect to external MCP servers if you need to. Almost all existing capabilities are preserved (except Computer Use, and marketplace)

This unlocks Claude Code for regulated industries. Banks, government agencies, defense contractors, healthcare, large telcos. Industries that were previously blocked from using the best coding agent can now adopt it.

Why This is Important for Solutions Architects

Enterprises hire the most amount of Solutions Architects. Over the next 6 to 12 months, every enterprise customer on AWS is going to ask the same question: "How do we adopt Claude Code securely?" If you can walk into that meeting and explain:

  • Why the Bedrock-native path exists
  • How to configure IAM and SSO for Claude Code access
  • How to think about cost allocation via Cost Explorer and tagging
  • How to layer in Bedrock Guardrails for prompt safety

Generalist SAs who understand this integration are going to be in high demand, because most architects still think of Claude Code as a "developer tool" and not as an enterprise workload.

This is exactly the kind of nuance that separates a good interview answer from a delightful one. When the interviewer asks, "How would you roll out a GenAI coding assistant at a regulated customer?", do not stop at "we use Claude Code." Go deeper. Talk about Bedrock-native inference, IAM-based access control, cost centralization, and MCP integration patterns. That is how you get hired.

Step by Step Setup

Here is how to wire it up. I just did this myself, and there are a couple of gotchas that the official docs do not call out clearly. I thought it'd be easier to show you, instead of writing. Check out the steps, in the below video:

video preview

Give this a try! With AWS investing $100B on Anthropic, I am confident that more and more features of Claude will be integrated with AWS.

Keep learning and keep rocking 🚀,

Raj

P.S - If you want to get an AWS Solutions Architect job without coding or learning every AWS service, the 8th cohort for AWS SA Bootcamp is launching on May 16th, 12 PM ET (Eastern Time) via live webinar. Please register for the webinar below:

Here’s what you get when you show up LIVE:

  1. My exclusive Solutions Architect framework to prep you for today's job market! But if you’re not live, you won’t get it. No second chances.
  2. Full bootcamp details, my new stealth product reveal to help you prepare better, AND a special offer for live participants only.
  3. You will have the chance to interact with me and ask questions.

And good news - it already worked for last cohort's students who secured cloud jobs in top companies, including at AWS, Microsoft, Google, JPMorgan, Reddit, and some of them didn't even have cloud experience 💰.

Spots are limited, so don't miss it!

Fast Track To Cloud

Free Cloud Interview Guide to crush your next interview. Plus, real-world answers for cloud interviews, and system design from a top AWS Solutions Architect.

Read more from Fast Track To Cloud

Hello Reader, I have interviewed hundreds of candidates as a Principal Solutions Architect at AWS, and helped many of my students crack top jobs. And here is what I keep seeing: candidates who are technically qualified still get rejected, because their answers sound exactly like every other candidate's. In today's edition, I will walk you through few of the most common Solutions Architect interview questions. For each one, I will share the average answer that quietly gets you rejected, and...

Hello Reader, SNS versus SQS versus EventBridge is one of the most common comparison questions I see, both in Solutions Architect interviews and in real customer conversations. A lot has changed on the features recently. In today's edition, I will level-set what these three services are, then compare them on every factor that matters. I have used these exact same factors when I designed world-scale applications as a Principal Solutions Architect at AWS. Let's get started. The Foundation &...

Hello Reader, Are you thinking about becoming an AWS SA? Most IT professionals trying to break into Solutions Architecture do the same thing: they stack certifications, grind through online courses on countless Cloud and Gen AI topics, and hope recruiters call. They think more credentials equals more interviews. But that only leads to wasted money and more rejection. What SA Bootcamp does is different. Instead of piling on paper credentials, we focus on exactly what interviewers at top...