We are excited about the latest release of Red Hat Ansible Automation Platform which includes new features that can be used to help address the common AI infrastructure challenges. Our goal is to guide IT teams, accelerate time to value and enhance the value of artificial intelligence (AI) initiatives that companies prioritize. We have identified 3 automation use cases that support the deployment, management and ongoing operations of AI in your organization.
Emerging AI-focused infrastructure challenges
Infrastructure readiness. With the rapid pace of change and the pressing need for innovation, organizations often perceive managing AI platforms and workloads to be complex, costly and time-consuming. In reality, AI infrastructure can be managed just like traditional infrastructure—consistently and through automation.
Real-time response to alerts and conditions. Teams spend considerable time and manual effort identifying and resolving automated error reports and performance issues. Once AI and observability tools trigger events, it is critical to provide an event-driven automated response and remediation to resolve them in a way that reduces downtime, as well as minimizes manual intervention and human error.
Confidence in acting on AI inferences. Organizations must design AI strategies that are trustworthy, reliable and aligned with established business policies and compliance regulations. IT operations teams aim to ensure that AI-empowered automated actions across AI infrastructure and AIOps function properly and align with intended and repeatable outcomes.
How can these areas be addressed?
Companies address these challenges through several approaches. We have started with 3 use cases: standardize AI infrastructure and operations, enable AIOps and enforce IT workflow policy.
Use case 1: Standardize AI infrastructure and operations
IT operations teams are responsible and accountable for managing and operating AI infrastructure at scale for the organization. These teams apply the same consistent, automated practices they use for existing infrastructure to AI environments. With automation, IT teams streamline the deployment, usage and scaling of AI infrastructure—accelerating time to value for developers consuming the AI platform. Let’s explore a specific example where Ansible Automation Platform adds value in this context.
An AI platform is an essential part of the infrastructure, as it equips developers with all the necessary tools and dependencies to start building and working with AI quickly. Operations teams provide AI frameworks so that end users in the organization can focus on building, training and deploying models—and not on the infrastructure itself. Without automation, handling the setup, configuration and maintenance of AI tools on your own can be complex, time-consuming and may require skills you haven’t yet developed. A ready-to-use, turnkey solution not only reduces set-up time and effort, but also accelerates your time to value.
To maintain consistency across your hybrid cloud environment, you need a platform that supports deployment wherever your data lives. Red Hat AI is our portfolio of products and services, including Red Hat Enterprise Linux AI (RHEL AI), Red Hat OpenShift AI and Red Hat AI Inference Server, that accelerates time to market and reduces the operational cost of delivering AI solutions across hybrid cloud environments. Red Hat AI enables efficient tuning of small, fit-for-purpose models with enterprise-relevant data and provides the flexibility to deploy wherever the data resides.
Install, configure and maintain AI infrastructure and operations: Red Hat AI
In this first use case, we provide instructions on how to provision and configure Red Hat AI as part of your infrastructure. We’ll also show how you can install, configure and maintain AI models that reside on Red Hat AI (in this case it’s via InstructLab, but could be another third-party open source model available through Red Hat AI). This process can require executing hundreds of manual commands. The good news is that Ansible Automation Platform simplifies this process while enabling consistency and repeatability.
What new features make this possible?
Two new Ansible Content Collections are now available:
- The redhat.ai certified collection provides supported modules to automate Red Hat AI and InstructLab activities.
- The infra.ai validated collection provides opinionated Ansible Roles to automate the provisioning of AI infrastructure, taking advantage of the redhat.ai certified collection.
Install, configure and maintain AI infrastructure: Red Hat AI
Use case 2: Enable AIOps
Once the AI infrastructure and contained models are in place, additional automation workflows can be integrated to utilize them. Even with automation, IT teams spend too much time diagnosing, manually responding to and remediating automated error reports. Customers want to shift from a reactive to proactive model for addressing manual operations, errors and performance issues. Let’s take a look at this specific example of how Ansible Automation Platform can help.
Closed-loop AI automation for Windows and Linux
Observability tools are already connected to Event-Driven Ansible, as well as into popular IT service management (ITSM) tools, so let’s now bring in generative AI (gen AI) to propose and then execute automated remediation. This reduces ticket resolution times, while accelerating IT productivity by automatically addressing logged events from multiple sources of truth. In this second use case, we provide instructions on how you can use self-healing infrastructure.
Logging and event monitoring are standard practice. In this example, an error takes place (example: Linux or Windows system service error) and is captured. As part of the Event-Driven Ansible feature of Ansible Automation Platform, rulebooks stand ready to receive generic events from external observability tools, such as ‘severity=1’. Ansible Automation Platform communicates with AI to interpret the error and provide a solution. After decision-making in the rulebook is complete, a playbook collects all the data and then sends it to gen AI for analysis and explanation. Ansible Automation Platform synchronizes the error details with ITSM (ServiceNow in this example) to create or update an incident. Available as an option for users to help address potential skills gaps, Red Hat Ansible Lightspeed can generate an Ansible Playbook based on the specific failure encountered. With our latest innovation—Lightspeed intelligent assistant (in technology preview) running on-premise, taking advantage of Red Hat AI—you can access trusted Ansible data sources directly through the platform’s chat interface, enabling faster onboarding, problem-solving and troubleshooting. Now, your operating system error is resolved!
Closed-loop AI automation for Windows and Linux
What new features make this possible?
- Event-Driven Ansible aggregates and simplifies the ecosystem of events and observability tools for AIOps workflows.
- Taking advantage of inferencing capabilities on Red Hat AI solutions to introspect events in Event-Driven Ansible and produce playbooks via Ansible Lightspeed.
- Red Hat Ansible Lightspeed intelligent assistant (tech preview) is an intuitive chat assistant embedded directly into the Ansible Automation Platform UI. It offers contextual assistance to accelerate troubleshooting, streamline platform onboarding and support daily automation management.
Use case 3: Enforce policies
How do you apply safeguards to the remediations you receive from AI? Now you can extend closed-loop AI automation to make sure that the automation abides by higher-order security, business and compliance rules. Let’s walk through a real example of how Ansible Automation Platform can help you keep AI in check as it rolls out across your stack.
Policy enforcement for AI
With Ansible Automation Platform, you can implement a policy enforcement approach to check policies at automation runtime, enabling you to intercept and prevent potentially harmful AI decisions before they occur—thus providing boundaries within which AIOps can run.
We begin with an AI-driven process that makes an inference from data, triggering automation. The AI then executes the automation content. As it runs, policy enforcement kicks in to scan and validate it against the configured policies. If it passes, the automation executes. If non-compliance with the policy is found, a human decision-maker is now involved to determine the next steps. This enables you to maintain control over AI-driven automation before it reaches completion, enhancing compliance, auditability and overall AI value. You can have confidence knowing that your event-driven automated responses are aligned with company policies.
Policy enforcement for AI
What new features make this possible?
- The policy enforcement feature integrates and enforces external policies at automation execution time across job templates, inventories, and Ansible organizations for better control and compliance tracking.
How can you get started with these use cases?
Ansible Automation Platform customers can find complete details about these use cases in Red Hat Customer Portal.
Not yet a customer? We encourage you to check out our Developer Program, and start an Ansible Automation Platform trial.
Additional resources:
Learn more at redhat.com/automation-ai
关于作者
Michele is an evangelist for automation, edge computing, artificial intelligence, and open hybrid cloud. She works closely with customers and partners to understand market requirements and then works with product management and engineering to ensure technology and solutions meet key market needs and will help move business value. She collaborates with partners to support strategic joint go-to-market initiatives in target verticals. Michele enjoys being an active member of global IoT consortiums, including the IoT Community, serving on the Board of Directors and also as the Co-Chair of the Women in IoT Center of Excellence since 2019. Prior to joining Red Hat in 2021, Michele spent seven years at SAS in various roles involving IoT and AI go-to-market strategy. She became one of the initial team members of the IoT Division, which she helped charter and in which she led product and partner marketing. Michele also held various roles during her 10-year tenure at IBM, including product management, product marketing, and program management.
