Claude Desktop With Bedrock: Compatibility and Options
Claude Desktop and AWS Bedrock are two separate ways to access Claude. They use different credentials, different interfaces, and different setups. This guide explains the differences, what each is for, and how fenbs keeps your work context when you move between clients.
Updated 4 min read
Claude Desktop is Anthropic's downloadable application, available for macOS and Windows, that signs in through a claude.ai account. AWS Bedrock is Amazon's managed AI service, which gives enterprise teams access to Claude and other models through AWS credentials and the AWS console. They are not interchangeable: Claude Desktop does not connect through Bedrock, and Bedrock access does not require or involve Claude Desktop.
If you use different clients across a team — some people on Claude Desktop, others on Bedrock-connected tools — fenbs gives every client the same shared work context over MCP, so switching client does not mean losing the project record.
What Claude Desktop is
Claude Desktop is a native app for macOS and Windows that provides a conversation interface to Claude. It signs in with a claude.ai account (Anthropic's own service), and its features and available Claude models depend on the plan attached to that account.
The key capability for tool workflows is MCP support: Claude Desktop can connect to external MCP servers, letting Claude read and update outside tools such as file systems, databases, and task boards. This is configured in a JSON file on your local machine. The Claude Desktop MCP config guide explains the setup.
What AWS Bedrock is
AWS Bedrock is Amazon's fully managed AI service that gives organizations access to foundation models — including Claude — through the AWS platform. It uses AWS credentials (IAM roles, access keys) rather than Anthropic account credentials. Access, billing, and data handling are all managed through AWS.
Bedrock is typically used by engineering teams that want to call Claude in code — building applications, running batch inference jobs, or integrating Claude into AWS-hosted services. It is not a desktop application; you call it through the AWS SDK or API.
Key differences at a glance
| Claude Desktop | AWS Bedrock | |
|---|---|---|
| What it is | A native desktop app for conversations with Claude | An AWS managed service for calling AI models in code |
| Credentials | claude.ai account (Anthropic) | AWS IAM credentials |
| Who uses it | Individuals and teams using Claude as an assistant | Engineering teams building with Claude via API |
| MCP support | Yes, via config file on the local machine | No built-in MCP; you add MCP support in your application code |
| Billing | Through Anthropic (claude.ai plans) | Through AWS |
| Claude model access | Models available on your claude.ai plan | Models available on Bedrock (check AWS docs for current list) |
Can you use Claude Desktop through Bedrock?
No. Claude Desktop connects to Anthropic's own service using a claude.ai account. There is no option in Claude Desktop to route traffic through Bedrock or authenticate with AWS credentials. If your organization requires all AI model calls to go through AWS, Claude Desktop is not the right tool.
For developers who want a terminal-based Claude experience that can be configured to use Bedrock, Claude Code supports Bedrock as a backend. The Claude Code with Bedrock guide explains the setup. The Claude Desktop vs Claude Code vs VS Code guide compares the available clients.
Using Bedrock with MCP
Bedrock does not include MCP support out of the box. If you are building an application on Bedrock and want it to call MCP servers, you add that capability in your application code: implement an MCP client, configure the servers, and handle the tool call loop yourself, or use a library that does this for you.
This is more work than using Claude Desktop, which handles the MCP client layer for you. For teams building on Bedrock who want Claude to read a shared work record, the options are to build a custom MCP client or use a Bedrock agent with action groups — check the current Bedrock documentation for supported patterns.
Keep the work when you move between clients
fenbs holds projects, tasks, Decisions and rules, and Where we left off outside any specific AI client. A team member using Claude Desktop and one using a Bedrock-connected tool read the same record when they connect over MCP. Switching client does not mean losing context. Free to start — connect from Integrations.
Create a free accountFree for one person · No credit card · Works with Claude, ChatGPT, Cursor and more
When to use each
Use Claude Desktop when you want a desktop assistant for day-to-day AI work, with MCP connections to your tools. It is the simplest way to connect Claude to fenbs, a file system server, or another tool without writing any code.
Use Bedrock when you are building an application that calls Claude in code, your organization requires AWS for AI model access, or you need enterprise features such as VPC endpoints, AWS CloudTrail logging, or specific data residency controls. Bedrock is an infrastructure choice, not a user-facing product. If both approaches are in use across a team, fenbs provides a neutral shared record that both Claude Desktop and Bedrock-connected apps can reach over MCP, so the model access layer does not also become the work context layer.
How to do this with fenbs
- Create a free account and set up your project in fenbs.
- If you use Claude Desktop: go to Integrations and follow the Claude Desktop setup guide. Connect with the browser sign-in flow.
- If you are building on Bedrock: implement an MCP client in your application and configure it to connect to fenbs at https://fenbs.ai/api/mcp. Sign in with a fenbs account to get the access token.
- Write the shared rules in Decisions and rules so every client reads them first, regardless of which model it runs.
- Use Where we left off on each project to hand context between clients and sessions.