Get With IT

Dave discusses the inefficiences of Kubernetes environments and the over-provisioning of resources with Debo Ray, co-founder and CEO of DevZero. Ray highlights the disconnect between engineers and finance departments, leading to unnecessary costs. He compares the cloud's short-term rental market to on-premises datacetners, noting the greater impact of over-provisioning in the cloud. Ray also discusses. the shift from CPU to GPU cycles in AI, emphasizing the need for better governance and visibility. Despite the rise of serverless infrastructure, Kubernetes remains a de facto standard for running applications, particularly in AI.

Dave Rubinstein, editor in chief, discusses updates to the Model Context Protocol (MCP) with Nate Barbettini, a founding member of Arcade.dev. Its solution acts as a layer between AI agents and systems, ensuring proper governance and authorization. The latest MCP release includes refinements to authorization to ensure consistency and correctness. The major update is MCP going stateless, making it more scalable and efficient, similar to HTTP. This change aims to support large-scale AI deployments, particularly in enterprises, despite the initial ecosystem disruption. 

Dave Rubinstein interviews Ilya Kabanov, a former Google AI security manager, about the disconnect between AI's rapid code generation and security measures. Kabanoff, who leads the Weather Report and is a research affiliate at MIT, discusses the significant progress AI has made in writing functional code, citing a 4x increase in security over a year. However, security checks cost five times more than code writing, posing economic challenges. Kabanov emphasizes the need for education, understanding AI's limitations, and prioritizing fundamental security practices to ensure secure AI-generated code.

Dave Rubinstein, Editor in Chief of IT Ops Times, discusses AI agentic architecture with Rodric Rabbah of Postman. Rabbah introduces Astro AI, a platform for managing AI agents, emphasizing the need for visibility and control in AI systems. He explains agentic architecture as a non-deterministic distributed system, highlighting the challenges organizations face in deploying and trusting AI. Rabbah stresses the importance of training, governance, and protocols to prevent AI from causing unintended consequences. He also notes the cost and practicality of using in-house models versus expensive external ones, advocating for a balance of AI, APIs, and data with robust guardrails.

Dave Rubinstein, Editor in Chief of IT Ops Times, discusses with Becky Trevino, Chief Product Officer at Flexera, the challenges organizations face in managing AI spending and ensuring value. Trevino explains the shift from traditional cloud and SaaS pricing to outcome-based pricing, driven by the rise of AI tools like ChatGPT and Claude. She highlights the need for organizations to revert to traditional metrics like cycle time and business outcomes to measure AI's effectiveness. Trevino emphasizes the importance of aligning AI investments with business goals and using tools like FinOps to manage costs effectively, comparing AI agents to temp workers that incur ongoing costs.

Dave Rubinstein and Christopher Robinson discussed the sustainability gap in open source package registries, which poses a threat to software supply chains. Robinson highlighted the exponential rise in compute costs and security demands, exacerbated by AI and malware proliferation. The Open Source Security Foundation formed a working group to address these issues, focusing on security technologies and community challenges. Robinson emphasized the need for responsible download practices and financial support for critical registries.

In this episode, Dave talks with Elad Koren, VP of Product Management at Palo Alto Networks, about how organizations rushing to implement AI without thinking about the security implications that can arise from that.

Among the topic they discuss are:

--The risk of proliferating unvetted agents

-- How vulnerabilities are being weaponized

-- What organizations can do to improve their security postures.

In this episode, editor-in-chief David Rubinstein discusses AI in the Security Operations Center with Jimmy Mesta, co-founder and CTO at RAD Security. Among the topics addressed are:

• How AI amplifies existing SOC strengths rather than inventing new capabilities from scratch?
• Why foundational elements, like signal hygiene, clear workflows, and documented context, must be in place before a SOC can benefit from AI.
• Lessons from past SIEM and SOAR deployments that security teams should apply to avoid repeating the same mistakes with AI in the SOC.
• The steps SOC leaders should take now (e.g., cleaning alerts, mapping workflows, documenting incidents) to ensure AI becomes leverage instead of liability.

In this episode, Dave interviews J.R. Storment, executive director of the FinOps Foundation about some of the challenges related to managing AI spend.

They discuss:

• The mismatch between increased spending on AI and a lack of clarity on that spend
• Similarities and differences with managing cloud spend
• How AI can actually be used for tracking AI spending

In this episode, Dave interviews Austin Parker, director of open source at Honeycomb and a member of OpenTelemetry's governance committee.

They discuss: 

• The increase in OpenTelemetry maintainers this year
• How prevalent OpenTelemetry has become across the cloud native ecosystem
• AI's impact on the need for observability