What’s Happening at Red Hat – February 2026

Hello Red Hat Enthusiasts

I recently reviewed Governor Newsom’s Executive Order N-12-23 on GenAI and mapped it to Red Hat’s latest GenAI e-book to highlight where they align. My goal is to give you a practical starting point whether you’re planning, piloting, or already deploying. If you’d like to jump to the e-book directly follow the link, but I do hope the newsletter will provide an opening context for next steps for a deeper dive.

What EO N-12-23 Asks Agencies to Operationalize

Guidelines for GenAI procurement, uses, and training—including high-risk scenarios and topics like safety, algorithmic discrimination, and data privacy.
An inventory of current high-risk GenAI uses, owned by a designated senior leader and maintained over time.
Approved “sandbox” environments for GenAI pilots, plus pilot projects that measure service improvements for Californians and productivity support for state employees.
Workforce enablement, including training that helps employees spot/mitigate hallucinations, inaccuracies, and bias—while enforcing privacy and applicable laws/policies.

Production Ready Considerations and Why They Map Well to EO N-12-23

Containers + Kubernetes for repeatability and control: Containers package model servers and dependencies for consistent rollout, while orchestration adds multitenancy, RBAC, policy, and quota enforcement. This is useful for agency “sandbox” environments and controlled pilots.
GenAIOps/MLOps to scale responsibly: Treat AI like software by standardizing toolchains, automate with CI/CD, and keep traceability across code, models, and data so pilots can mature into governed production services.
Hybrid cloud foundations for data locality and sovereignty: A consistent hybrid platform helps keep sensitive data/models in controlled environments while still using cloud scale when appropriate—supporting the data residency/privacy expectations that show up in state guidance.
AI safety as a core requirement: The e-book calls out guardrails (policy controls, content filters, and tool-execution controls), plus continuous validation to detect drift, bias, and reliability issues that are directly aligned to the EO’s emphasis on safe and responsible GenAI use.

From Pilot to Production: the e-book’s 8-step AI platform checklist

Building gen AI powered applications and AI agents is an iterative process that extends beyond simply creating AI models. The main steps outlined in the Red Hat AI e-book are:

Define your use case, set business goals for your AI initiative, and get buy-in from stakeholders and leadership.
Choose where you want your model experimentation and deployment platforms to run: on premise or in the cloud.
Choose the AI model that best fits your needs. Avoid lock-in by choosing open models.
Customize or align your chosen models with your proprietary data using retrieval-augmented generation (RAG).
Deploy your model in an inference server.
Build gen AI-powered applications or workloads.
Once you have a working environment in place, extend and automate the workflow through agentic AI.
Monitor and manage models in a security-focused manner and at scale.

How Red Hat Can Help California Agencies Take the Next Step

If you’re planning (or already running) GenAI pilots, we can help with a practical “GenAI readiness” discussion that covers: