AWS Bedrock integration
The AWS Bedrock integration allows Catalyst workflows to use multiple AI providers through Amazon Bedrock.
Supported providers may include:
- Amazon Nova
- Anthropic Claude
- Meta Llama
- Mistral AI
This allows organisations to choose different AI models for different workflow requirements.
Typical use cases
AWS Bedrock models can be used for:
- contract analysis
- document summarisation
- legal research
- AI-assisted review
- compliance analysis
- document categorisation
- workflow automation
- risk assessment
Configurable AI parameters
The AWS Bedrock actor supports configuration of:
- temperature
- Top P
- Top K
- maximum length
- stop sequences
These settings help control:
- creativity
- predictability
- response length
- formatting consistency
- output structure
Choosing the right model
Different models may be better suited to different tasks.
Examples include:
- Claude for factual consistency
- Nova for contract analysis
- Llama for reasoning tasks
- Mistral for high-speed document processing
Things to remember
- Available models depend on your AWS Bedrock configuration.
- Different models may produce different outputs for the same prompt.
- Lower temperature settings typically produce more predictable responses.
- AI outputs should be reviewed before operational use.