SUSE Launches New AI Capabilities at SUSECON

SUSE is excited to announce several new AI capabilities for the SUSE AI platform at SUSECON. SUSE AI is designed to be the platform of choice when deploying and running AI workloads. These new features focus on observability, agentic workflows, guardrails, and expanding our AI library.

Observability for AI Workloads

AI Observability is the ability to monitor, understand, and debug AI systems in real time by collecting and analyzing relevant metrics and logs. Key components include:

• Operational Observability for tracking model uptime, latency, and overall system health.
• Model Observability for monitoring model performance, fairness, and interpretability.
• Data Observability for ensuring data quality, detecting anomalies, and tracking data lineage.
• GPU Monitoring provides insight into the health and use of a very expensive resource.

As AI transitions into production, understanding its performance and costs becomes essential. The SUSE AI preconfigured observability dashboards for AI workloads provide critical metrics on:

• LLM cost and token usage
• LLM performance
• LLM drift
• GPU performance
• VectorDB performance

Automatic instrumentation for Python, Java, and Go images enables data collection on model performance, latency, and failures without requiring code modification. This is similar to a GPS tracker in a car, where the car’s operation isn’t changed, but its performance and drift can be monitored.

With SUSE AI Observability, users can stay compliant with AI regulations, trust the output of their GenAI applications, optimize performance, empower developers with an observable and transparent process, and provide cost predictability through token usage and GPU performance

Enabling Agentic Workflows

Agentic workflows provide a way to simplify complex business problems and allow AI systems to operate autonomously, adapt to dynamic environments, and make decisions based on evolving inputs.

SUSE is providing tools to help businesses build agentic workflows, including a greendoc that offers a step-by-step reference architecture based on Langflow. 

Additionally, OpenWebUI Pipelines will be added to the AI Library. OpenWebUI Pipelines is an open-source framework that allows users to create and manage agentic workflows by chaining together AI models, APIs, and external tools.

Agentic workflows can plan, reason, and take actions based on goals and data. These workflows can create autonomous agents to execute tasks, make decisions, and adapt to dynamic environments without constant human intervention, unlocking unprecedented levels of automation and efficiency. AI agents can handle routine inquiries, provide personalized recommendations, qualify leads, schedule appointments, personalize pitches, diagnose illnesses, ask intelligent follow-up questions, and make doctor’s appointments.

Implementing Guardrails Technology

To address concerns about trust, ethics, and transparency, SUSE is enabling customers to build guardrails into their AI workloads. Guardrails prevent unintended behavior, biases, security risks, and compliance violations, ensuring AI outputs are safe, accurate, and aligned with business objectives.

A greendoc will guide users through implementing guardrail technology using the open-source Guardrails AI project. SUSE is also partnering with Infosys to integrate their Infosys Responsible AI with SUSE AI, providing enhanced ethical AI practices, data privacy, and regulatory compliance. Infosys Guardrails combines enhanced ethical AI practices with the SUSE AI platform’s security and observability, ensuring data privacy, regulatory compliance, and insight into AI workloads.

SUSE AI provides a reliable and compliant framework for AI workloads using SUSE Security. SUSE AI is secure by design, with AI compliance and governance, zero-trust protection, lifecycle coverage, limited access to approved model repositories, and an auditable framework to meet regulatory requirements.

Expanding the AI Library

The AI landscape is constantly evolving, so SUSE is expanding its AI Library to provide customers with access to secure and innovative open-source AI components. The AI Library will include PyTorch, OpenWebUI Pipelines, and MLflow in a rolling release format.

The SUSE AI Library contains curated AI tools and components built using the SUSE secure supply chain.

The SUSE AI Library includes:

• Ollama: For LLM management and deployment
• Milvus: A vector database tailored for similarity search on massive datasets
• Open WebUI: An extensible, feature-rich, self-hosted AI platform
• OpenWebUI Pipelines: Enables MLOps use cases and provides a platform for managing machine learning workflows
• MLflow: Enables MLOps use cases and provides a platform for managing machine learning workflows
• Pytorch: Enables a variety of use cases, such as computer vision, NLP, image classification, and tensor computing

Expanded Partnerships

SUSE is expanding partnerships with companies like Infosys and HPE to deliver complete enterprise AI infrastructure. SUSE, Infosys, and HPE deliver complete Enterprise AI Infrastructure. Infosys Responsible AI is integrated with SUSE AI.

These new capabilities demonstrate SUSE’s commitment to providing a comprehensive and secure AI platform.