AWS Bedrock Configuration
Required Fields
Region *
The AWS Region where your Bedrock service is deployed. Example:us-east-1
Common regions with Bedrock availability:
us-east-1(N. Virginia)us-west-2(Oregon)eu-west-1(Ireland)ap-northeast-1(Tokyo)ap-southeast-1(Singapore)
- Select a region close to your users for lower latency
- Ensure the models you need are available in that region
- Consider data residency requirements for your organization
Model Name *
The Model Name field specifies the model ID or ARN you want to use with Novaplan AI. Model ID format examples:us.anthropic.claude-sonnet-4-20250514-v1:0anthropic.claude-3-5-sonnet-20241022-v2:0amazon.titan-text-express-v1meta.llama3-70b-instruct-v1:0
arn:aws:bedrock:us-east-1:106782021127:inference-profile/apac.anthropic.claude-sonnet-4-20250514-v1:0
- Go to the AWS Bedrock Console
- Navigate to “Model access” to see available models
- Select “Base models” or “Custom models” to view model IDs
- Use the model ID as shown in the console
Provider *
Select the provider of the foundation model you’re using. This helps Novaplan AI properly format requests for the specific model. Available providers:- Anthropic (Claude) - Claude models
- Mistral - Mistral and Mixtral models
- Qwen - Alibaba’s Qwen models
- DeepSeek - DeepSeek models
- Cohere - Cohere Command models
- Amazon (Titan) - Amazon’s Titan models
- AI21 Labs - AI21 Labs models
- Other - For any other providers (requires Custom Provider Name)
Custom Provider Name
Only required if you selected “Other” as the provider. Enter the name of the custom provider you’re using.Optional Fields
AWS Access Key ID
The AWS Access Key ID for authenticating requests to AWS Bedrock. Example format:AKIAIOSFODNN7EXAMPLE
Both AWS key fields can remain blank when Novaplan has configured an approved credential source for the managed backend, such as a workload role. The accessible backend then uses the Boto3 credential provider chain; an AWS login on your own computer does not automatically give the Novaplan service access. Confirm the deployed workspace’s credential source with Novaplan before saving this configuration.
AWS Access Secret Key
The secret portion of your AWS credentials. Use an explicit access key pair only when your AWS administrator and Novaplan have agreed that the managed deployment cannot use an approved role or another temporary-credential method. Ask them to define the required IAM permissions, secret-transfer method, and rotation plan. Never put a real key in documentation, a ticket, or a repository. AWS recommends temporary credentials where possible.Context Length
Specify the maximum context length for your model (e.g., 128000 for 128K tokens). This helps Novaplan AI optimize how it handles large documents and conversations. Note: Different models support different context lengths. Check your model’s documentation for the maximum supported context.Multimodal
Enable this option if your model supports multimodal inputs (text + image). This allows Novaplan AI to process both text and images when interacting with the model. Models with multimodal support include:- Claude 3.5 Sonnet, Claude 3 Opus, Claude 3 Haiku
- Amazon Titan Image models
Reasoning
Enable this option if your model supports advanced reasoning capabilities. This is particularly useful for models designed for complex problem-solving tasks.Configuration Steps
As shown in the image above:- Click Configure on the Amazon Bedrock provider card
- Specify your AWS Region (marked with *)
- Enter the Model Name (model ID or ARN) (marked with *)
- Select the appropriate Provider from the dropdown (marked with *)
- If you selected “Other” as provider, enter the Custom Provider Name
- (Optional) Enter the approved AWS Access Key ID only if Novaplan is using explicit keys
- (Optional) Enter its matching AWS Access Secret Key through the approved setup flow
- (Optional) Set the Context Length for your model
- (Optional) Enable Multimodal if your model supports text + image inputs
- (Optional) Enable Reasoning if your model supports advanced reasoning
- Click Add Model to save and validate your credentials
Region, Model Name, and Provider are required. The two key fields may be blank only when the managed backend has another approved AWS credential source.
IAM Permissions
The AWS principal used by Novaplan needsbedrock:InvokeModel for model inference and bedrock:InvokeModelWithResponseStream if the chosen call path streams responses. An inference profile or a different Bedrock API path may need additional permissions and resource ARNs. Have your AWS administrator scope the policy to the model or profile used by this workspace rather than copying an unrestricted Resource: "*" example. See AWS’s inference prerequisites and IAM policy examples.
Usage Considerations
- API usage will count against your AWS Bedrock quotas and billing
- Different models have different pricing - check AWS Bedrock pricing for details
- Model capabilities vary - more powerful models may provide better results but at higher cost
- AWS Bedrock provides enterprise features including:
- VPC endpoints for private connectivity
- AWS PrivateLink support
- CloudWatch integration for monitoring
- CloudTrail for auditing
- Data encryption at rest and in transit
Troubleshooting
- If you encounter authentication errors, verify the AWS identity configured for the managed backend or, if explicit keys were approved, the matching key pair
- Ensure that AWS identity has the necessary Bedrock permissions
- Verify the selected region has AWS Bedrock available and your desired model is accessible
- Check that you have model access enabled for the specific model in the AWS Bedrock console
- Ensure the model ID or ARN is correctly formatted
- If using an inference profile, verify the ARN is complete and accurate
- Check your AWS account’s service quotas for Bedrock