SAS Retrieval Agent Manager Features List
No-code/low-code user experience
- Reduces dependency on developers for coding with drag-and-drop automation, enabling faster adoption of complex workflows
- Reduces time-to-value by enabling rapid prototyping and iteration of agentic workflows.
Flexible AI stack integration
- Prevents vendor lock-in via plugin-based integrations, modular design and open standards.
- Supports leading model providers, including OpenAI, Azure OpenAI, Amazon Bedrock and Ollama.
- Integrates with popular vector databases such as PGVector and Weaviate.
Broad data support compatibilities
Optimized RAG performance
- Reduces memory footprint and latency via model compression for RAG workflows.
- Boosts inference speed with quantization while maintaining relevance and accuracy.
- ONNX acceleration boosts performance, enabling faster, more efficient inference across diverse hardware.
Trustworthy responses and human-in-the-loop oversight
- Provides built-in automated and user-driven evaluations to identify the optimal configuration for reliable responses.
- Auto-generates citations linking answers to source documents, ensuring transparency and traceability.
- Enhances trust and regulatory compliance with explainable and auditable outputs.
- Supports human-in-the-loop validation for critical RAG decisions.
Versatile deployment options
- Supports on-premises and air-gapped deployments, giving organizations full control over their data in highly secure or isolated environments.
- Deployable across public and private clouds, enabling flexibility, scalability and alignment with enterprise IT strategies.
Fine-grained security & privacy
- Provides granular role-based access and privacy controls from source documents to agents to ensure secure handling of sensitive information.
- Ensures comprehensive logging, auditing and traceability across the RAG workflow to support compliance and maintain transparency.
Extensible agentic automation
- Supports building, deploying and managing retrieval agents through a plugin-based architecture that seamlessly integrates external APIs, tools and enterprise services.
- Autonomous agent orchestration enables agents to retrieve, reason and act across systems, automating complex workflows with contextual awareness, precision and scalability.
Model context protocol (MCP) empowered
- MCP integration enables agents to move beyond retrieval and trigger real business processes, APIs and multistep tasks with structured tool calls.
- Transparent input and output schemas for every tool call, along with full observability, ensure safe, auditable and compliant AI actions.
- Standardized, reusable tool servers encapsulate enterprise logic, reducing integration costs and accelerating deployment across teams.
- Structured tool invocation minimizes token usage, improves speed and delivers deterministic outputs.