Software supply chain security presents an architecture problem. Sudeep Goswami of Traefik Labs appears on AppDevANGLE, produced by theCUBE Research, to discuss Traefik Labs' role in cloud native proxy and gateway solutions and the company's Distro Zero initiative. Goswami explains their team's experience with Kubernetes deployments and describes the architectural shift from vulnerability detection toward attack surface reduction across application programming interface, artificial intelligence and MCP gateway workloads. They outline how Distro Zero reduces the attack surface by shipping a minimal memory-safe binary and by simplifying software bill of materials management, SBOM, while easing Federal Information Processing Standard 140-3, FIPS 140-3 and broader compliance efforts.Paul Nashawaty of theCUBE Research reports that 58% of respondents rely on vulnerability scanning while 47% list software supply chain security as a top investment priority. Nashawaty adds that 54% cite the National Institute of Standards and Technology framework, NIST and 46% cite the EU Cyber Resilience Act.Goswami emphasizes technical and operational implications for cloud native architectures, Kubernetes deployments and gateway workloads, including API gateway, AI gateway and MCP use cases. They discuss best practices for supply chain security, vulnerability management, SBOM handling and compliance in enterprise environments.

Susan Laine of Quest Software, global field chief technology officer CTO, discusses AI governance and data readiness on AppDevANGLE. This episode focuses on governance and trusted data in enterprise artificial intelligence AI deployments. Laine brings practical experience in data management and modernization and they outline challenges related to data readiness, semantic layers, model retraining and operationalizing AI. Hosts from AppDevANGLE and Paul Nashawaty of theCUBE Research contribute additional perspective.Laine explains how metadata, data products and a control plane factor into moving validated models safely into production while mitigating risks from agentic AI behavior. Key takeaways include a widening governance gap and the need for durable, change-ready frameworks. Paul Nashawaty of theCUBE Research observes that many organizations rely on public AI tools while lacking enterprise-wide governed frameworks. Laine recommends building trust through semantic layers, reusable data products and a control plane and they suggest using AI for AI to scale safe data modeling and maintain production reliability. The discussion highlights implications for data governance, data strategy, machine learning operations MLOps and artificial intelligence operations AIOps in enterprise AI modernization.

This episode examines how artificial intelligence, abbreviated as AI, agents transform software delivery to accelerate developer productivity and improve developer experience. Paul Nashawaty of theCUBE Research hosts the discussion with Deepak Singh of AWS and Steve Charza of Amazon.com. The conversation explores Kiro's capabilities, spec-driven workflows, agent steering, context as code and real-world applications of agentic automation to scale engineering.Nashawaty reports research showing 65% of respondents spend up to 20% of engineering time on net new innovation, highlighting a substantial productivity gap. Singh observes that teams that redesign workflows and give agents clear intent and context see outsized gains; they recommend spec-driven approaches and agent steering to capture those gains. Charza reports median shipping speed improvements of 4.5× and cites spec-driven development combined with Kiro and agentic automation as enabling features to ship months faster.

This episode examines Meko and shared memory for agentic artificial intelligence knowledge infrastructure. The guest, co-founder and CEO of Yugabyte, draws on a decade of experience in distributed systems, databases and cloud-native architecture. In conversation with Paul Nashawaty of theCUBE Research on AppDev ANGLE, they explore Yugabyte's new product Meko, agent-native data infrastructure, shared memory for agent collaboration and how data architecture underpins agentic AI deployments and enterprise use cases.The co-founder and CEO of Yugabyte explains that the primary limitation on agentic AI is data infrastructure rather than models and that shared observable memory across agents reduces token costs and operational complexity. They describe how Meko delivers auditability, multi-tenant scalability and efficiency while promoting collective memories and traceable reasoning to support compliance, return on investment and more predictable enterprise agent workflows. The discussion highlights implications for observability, vector databases, PostgreSQL compatibility and open source strategies for enterprise adoption.Watch to understand the role of data architecture in scalable agent deployments and practical factors to consider when implementing agent-native knowledge infrastructure in production environments.

This discussion examines identity security for artificial intelligence agents and the role of zero trust in protecting enterprise environments. Geoff Mattson of SecureAuth, chief executive officer, joins Paul Nashawaty of theCUBE Research, practice lead and principal analyst, to examine how SecureAuth addresses agentic threats as AI agents expand the enterprise attack surface. Mattson highlights passkeys, behavioral analytics and continuous identity evaluation as core components, and they describe combining static authorization policies with behavioral profiling and analytics driven step up and step down authentication to reduce phishing, minimize user friction and mitigate rogue agent behavior. Nashawaty underscores rising attack volumes and recommends identity first zero trust architectures and real time controls.Key takeaways include treating agents as dynamic identities that require transaction level evaluation, according to Mattson. They emphasize that combining static policies with behavioral profiling and continuous identity evaluation reduces risk while preserving user experience. Nashawaty and theCUBE Research advocate identity first zero trust, real time controls and stronger authentication models to manage agentic threats across finance, healthcare and retail. This conversation addresses authentication, authorization and identity and access management within modern cybersecurity strategies.

The next generation of AI-powered commerce is moving beyond recommendation engines toward intelligent reasoning. In this episode of AppDevANGLE, Paul Nashawaty speaks with Konstantin Kiselev, CTO and Co-Founder of Haut.AI, about how enterprises are using causal reasoning, knowledge graphs, computer vision, and clinical validation to build trusted AI experiences that go far beyond traditional product recommendations. The discussion explores why statistical correlation alone is insufficient for high-consequence decisions, how domain-specific reasoning improves consumer trust, and why explainable AI is becoming a strategic requirement for enterprise applications in healthcare, beauty, and personalized commerce. As organizations increasingly adopt agentic AI, the ability to combine structured knowledge with real-world validation may become a key differentiator for next-generation customer experiences. Key Highlights: The evolution from recommendation engines to reasoning-based AI Why explainable AI is critical for personalized commerce How causal knowledge graphs improve AI decision-making Clinical validation as a foundation for trusted enterprise AI Continuous feedback loops and adaptive AI recommendations The importance of domain intelligence in agentic AI systems Building AI applications that balance personalization with trust Enterprise implications for the future of agentic commerce

Brian Fox of Sonatype, co-founder and chief technology officer, joins Paul Nashawaty of theCUBE Research, practice lead and principal analyst and host of AppDevANGLE, to examine the intersection of artificial intelligence and software supply chain security.Fox draws on open source experience and stewardship of Maven Central to discuss dependency management, the limits of ungrounded large language models and retrieval-augmented generation and the need to encode policies so AI can make secure dependency choices at scale. They explain that AI accelerates code production but can amplify vulnerability debt when models lack real-time supply chain data. They recommend integrating live feeds and automated dependency governance so AI reasons from accurate facts. Nashawaty emphasizes that consistent security controls and machine-scale automation are required to prevent persistent false negatives and enable timely remediation across thousands of applications.This discussion provides practical guidance for DevSecOps teams on integrating real-time supply chain telemetry into development workflows, improving dependency management and reducing vulnerability exposure in production.

This conversation addresses enterprise data fragmentation governance and artificial intelligence AI return on investment ROI challenges. Sam Newnam of Hammerspace outlines the company’s AI data platform and describes work on packaging AI solutions original equipment manufacturer OEM and cloud partnerships and on data orchestration. Paul Nashawaty of theCUBE Research hosts the AppDev ANGLE podcast.Newnam explains strategies to address hybrid and multi cloud data fragmentation the enterprise skills gap metadata-centric governance and practical approaches for preparing unstructured data to accelerate AI initiatives into production. They emphasize that data rather than models or graphics processing units GPU is the primary barrier to AI ROI. Nashawaty estimates that 60 to 80 percent of AI initiatives never reach production because of fragmented ungoverned data. Newnam recommends an AI data platform that centralizes metadata enforces policy-driven orchestration and enables project-based pilots to reduce friction close the skills gap and scale outcomes.Listen for actionable strategies on metadata governance data orchestration hybrid cloud and accelerating AI adoption in the enterprise.

Peter Farkas of Percona joins theCUBE Research host Paul Nashawaty of AppDev to discuss running stateful databases on Kubernetes, the evolution of operators and Percona's approach to open-source data platforms. Farkas draws on Percona's two decades of database expertise across MySQL, PostgreSQL, MongoDB and Valkey and explores platform integration, operator maturity and practical deployment trade-offs. They address storage considerations, operator-driven automation and cloud native deployment patterns.Key takeaways include that Kubernetes now supports many production database workloads through mature operators and storage layers, though ultra-low-latency or regulatorily isolated cases may still require bare metal, Farkas observes. They advocate augmented artificial intelligence rather than fully autonomous systems for database management and emphasize human-in-the-loop approaches combined with practical automation. theCUBE Research highlights operational efficiencies and the importance of vendor-agnostic community-driven open-source tooling and outcome-focused service bundles.

Molham Aref of RelationalAI joins Paul Nashawaty of theCUBE Research, practice lead and principal analyst, on AppDev ANGLE to discuss how enterprises bridge the gap between model capability and realized business value through relational context and executable semantics. The discussion focuses on practical approaches for deploying artificial intelligence, AI, in production by embedding business semantics into data architectures to move from isolated copilots to trusted production-grade systems. Aref explains how relational context and executable semantics reduce token waste and enable efficient tooling for prediction, prescription and graph reasoning. Nashawaty emphasizes return on investment, ROI and operational scalability and they highlight that combining smaller models with post-training on business-specific data can lower costs while improving decision quality.Key takeaways include the necessity of relational and executable context rather than solely textual context. Embedding business semantics into data architectures supports prediction, prescription and graph reasoning and reduces computational and token inefficiencies. Adopting business-aware data architectures and governance enables trusted production-grade AI and improves ROI. Operational strategies include combining smaller foundational models with post-training on business-specific data, implementing machine learning operations, MLOps and focusing on cost optimization and scalable deployment.