Agentic AI - How to with SAS® Viya®
AGAI : AGAIV6
Learn to build, deploy, and monitor Agentic AI LLM-based applications using SAS Viya and the SAS Agentic AI Accelerator. You will also learn how to expose SAS Viya capabilities through the SAS Viya MCP Server so AI coding clients such as Claude Code and GitHub Copilot can discover tools, execute SAS tasks, and guide multi-step workflows from VS Code.
Learn How To
- Register, publish, deploy, and monitor large language models (LLMs) and Agentic AI workflows (intelligent decision flows).
- Combine proprietary and open-source LLMs with deterministic models in decision-making workflows.
- Govern, version, and scale LLM usage in enterprise applications.
- Configure and use the SAS Viya MCP Server with Claude Code and GitHub Copilot from VS Code.
Who Should Attend
- Technical Roles:
- Data Scientists and AI Engineers
- Cloud Architects
- Developers
- Business Roles:
- Decision Science Professionals
- Mixed Roles:
- Enterprise Application Teams
- AI Strategists
Prerequisites
- SAS Model Manager and ModelOps.
- SAS Intelligent Decisioning.
- SAS Container Runtime.
SAS Products Covered
SAS Model Manager;SAS Intelligent Decisioning
Course Outline
Foundational Model Repository
- Introduction to SAS Agentic AI Accelerator Model Repository.
- Standardizing inputs, outputs, and options using Python wrappers.
- Register proprietary and open-source models in SAS Model Manager.
- Publish models as Docker images to Azure Container Registry.
- Learn to deploy LLMs to Kubernetes, focusing on secure HTTPS endpoints.
- Deploy models to Azure containers using scripts.
- Score deployed models using SAS, Python, or bash scripts.
- Configure security settings for CORS, CSRF, cookies, and Content Security Policy (CSP). Set up Prompt Builder configuration using SAS Visual Analytics.
- Create and run prompt experiments with LLMs, save and manifest the best prompt, and deploy it in SAS Model Manager.
- Integrate your prompt with SAS Intelligent Decisioning and validate the workflow.
- Create decision flows with non-deterministic LLMs (e.g., GPT, PHI) and deterministic models (e.g., gradient boosting, sentiment analysis).
- Modify LLM configurations, update prompts, and version workflows.
- Publish the Agentic AI decision to SAS Micro Analytic Service as a module and score it in real time over a REST API, with no separate deployment step.
- Consume the deployed agent from enterprise applications and AI coding clients through its SAS Container Runtime / MAS REST endpoint.
- Learn to collect logs, parse them, and populate a dashboard to monitor LLM usage, performance, responses, and sentiment.
- Configure the SAS Viya MCP Server for OAuth callback, local runtime, and VS Code connectivity.
- Connect Claude Code and GitHub Copilot to SAS Viya through MCP.
- Use MCP tools to discover published models and MAS modules, score decisions, and run multi-step SAS and CAS workflows.
Live Class Schedule
Duration: 14 hours
Step into our live classes and experience a dynamic learning environment where you can ask questions, share ideas, and connect with your instructor and classmates. With on-demand lab hours, you can explore the material at your own pace. Our globally acclaimed instructors will motivate you to think bigger, so you can take what you've learned and achieve your biggest goals.
This course isn't publicly scheduled, but private training and coaching may be available. Contact us to explore options.
Private Training
Get training tailored specifically for your team, led by expert SAS instructors. Choose from virtual sessions, or training at your location (or ours). Perfect for teams seeking a customized curriculum and plenty of interaction with a SAS specialist. We'll schedule it at a time that works for you.
Coaching Services
Take your training to the next level with personalized coaching. While private training offers structured coursework, coaching provides hands-on, real-time support from a subject matter expert. As you work with your own data, you'll receive expert guidance to help you uncover insights, unlock the full potential of your data, and make faster progress. Perfect for those looking to apply what they’ve learned and see quicker results.