The New Stack Podcast

Woodson Martin, CEO ofOutSystems, argues that successful enterprise AI deployments rarely rely on standalone agents. Instead, production systems combine AI agents with data, workflows, APIs, applications, and human oversight. While claims that “95% of agent pilots fail” are common, Martin suggests many of those pilots were simply low-commitment experiments made possible by the low cost of testing AI. Enterprises that succeed typically keep humans in the loop, at least initially, to review recommendations and maintain control over decisions. Current enterprise use cases for agents include document processing, decision support, and personalized outputs. When integrated into broader systems, these applications can deliver measurable productivity gains. For example,Travel Essencebuilt an agentic system that reduced a two-hour customer planning process to three minutes, allowing staff to focus more on sales and helping drive 20% top-line growth. Martin also believes AI will pressure traditional SaaS seat-based pricing and accelerate custom software development. In this environment, governed platforms like OutSystems can help enterprises adopt “vibe coding” while maintaining compliance, security, and lifecycle management. Learn more from The New Stack about the latest developments around enterprise adoption of vibe coding: How To Use Vibe Coding Safely in the Enterprise 5 Challenges With Vibe Coding for Enterprises  Vibe Coding: The Shadow IT Problem No One Saw Coming Join our community of newsletter subscribers to stay on top of the news and at the top of your game. 

On a recent episode of the The New Stack Agents, Inception Labs CEO Stefano Ermon introduced Mercury 2, a large language model built on diffusion rather than the standard autoregressive approach. Traditional LLMs generate text token by token from left to right, which Ermon describes as “fancy autocomplete.” In contrast, diffusion models begin with a rough draft and refine it in parallel, similar to image systems like Stable Diffusion. This parallel process allows Mercury 2 to produce over 1,000 tokens per second—five to ten times faster than optimized models from labs such as OpenAI, Anthropic, and Google, according to company tests. Ermon argues diffusion models better leverage GPUs, with support from investor Nvidia to optimize performance. While Mercury 2 matches mid-tier models like Claude Haiku and Google Flash rather than top systems such as Claude Opus or GPT-4, Ermon believes diffusion’s speed and economic advantages will become increasingly compelling as AI applications scale. Learn more from The New Stack about the latest developments around around large language model built on diffusion:  How Diffusion-Based LLM AI Speeds Up Reasoning Get Ready for Faster Text Generation With Diffusion LLMs  Join our community of newsletter subscribers to stay on top of the news and at the top of your game.   

OnThe New Stack Agents, Gavriel Cohen discusses why he built NanoClaw, a minimalist alternative to OpenClaw, after discovering security and architectural flaws in the rapidly growing agentic framework. Cohen, co-founder of AI marketing agencyQwibit, had been running agents across operations, sales, and research usingClaude Code. When Clawdbot (laterOpenClaw) launched, it initially seemed ideal. But Cohen grew concerned after noticing questionable dependencies—including his own outdated GitHub package—excessive WhatsApp data storage, a massive AI-generated codebase nearing 400,000 lines, and a lack of OS-level isolation between agents. In response, he createdNanoClawwith radical minimalism: only a few hundred core lines, minimal dependencies, and containerized agents. Built around Claude Code “skills,” NanoClaw enables modular, build-time integrations while keeping the runtime small enough to audit easily. Cohen argues AI changes coding norms—favoring duplication over DRY, relaxing strict file limits, and treating code as disposable. His goal is simple, secure infrastructure that enterprises can fully understand and trust.   Learn more from The New Stack about the latest around personal AI agents Anthropic: You can still use your Claude accounts to run OpenClaw, NanoClaw and Co. It took a researcher fewer than 2 hours to hijack OpenClaw OpenClaw is being called a security “Dumpster fire,” but there is a way to stay safe Join our community of newsletter subscribers to stay on top of the news and at the top of your game. 

A few weeks after Dynatrace acquired DevCycle, Michael Beemer and Andrew Norris discussed on The New Stack Makers podcast how feature flagging is becoming a critical safeguard in the AI era. By integrating DevCycle’s feature flagging into the Dynatrace observability platform, the combined solution delivers a “360-degree view” of software performance at the feature level. This closes a key visibility gap, enabling teams to see exactly how individual features affect systems in production. As “agentic development” accelerates—where AI agents rapidly generate code—feature flags act as a safety net. They allow teams to test, control, and roll back AI-generated changes in live environments, keeping a human in the loop before full releases. This reduces risk while speeding enterprise adoption of AI tools. The discussion also highlighted support for the Cloud Native Computing Foundation’s OpenFeature standard to avoid vendor lock-in. Ultimately, developers are evolving into “conductors,” orchestrating AI agents with feature flags as their baton.   Learn more from The New Stack about the latest around AI enterprise development:  Why You Can't Build AI Without Progressive Delivery  Beyond automation: Dynatrace unveils agentic AI that fixes problems on its own  Join our community of newsletter subscribers to stay on top of the news and at the top of your game.   

Dynatrace is at a pivotal point, expanding beyond traditional observability into a platform designed for autonomous operations and security powered by agentic AI. In an interview on *The New Stack Makers*, recorded at the Dynatrace Perform conference, Chief Technology Strategist Alois Reitbauer discussed his vision for AI-managed production environments. The conversation followed Dynatrace’s acquisition of DevCycle, a feature-management platform. Reitbauer highlighted feature flags—long used in software development—as a critical safety mechanism in the age of agentic AI. Rather than allowing AI agents to rewrite and deploy code, Dynatrace envisions them operating within guardrails by adjusting configuration settings through feature flags. This approach limits risk while enabling faster, automated decision-making. Customers, Reitbauer noted, are increasingly comfortable with AI handling defined tasks under constraints, but not with agents making sweeping, unsupervised changes. By combining AI with controlled configuration tools, Dynatrace aims to create a safer path toward truly autonomous operations. Learn more from The New Stack about the latest in progressive delivery: Why You Can’t Build AI Without Progressive Delivery Continuous Delivery: Gold Standard for Software Development Join our community of newsletter subscribers to stay on top of the news and at the top of your game. 

Matan-Paul Shetrit, Director of Product Management at Writer, argues that people must take responsibility for how they use AI. If someone produces poor-quality output, he says, the blame lies with the user—not the tool. He believes many misunderstand AI’s role, confusing its ability to accelerate work with an abdication of accountability. Speaking on The New Stack Agents podcast, Shetrit emphasized that “we’re all becoming editors,” meaning professionals increasingly review and refine AI-generated content rather than create everything from scratch. However, ultimate responsibility remains human. If an AI-generated presentation contains errors, the presenter—not the AI—is accountable. Shetrit also discussed the evolving AI landscape, contrasting massive general-purpose models from companies like OpenAI and Google with smaller, specialized models. At Writer, the focus is on enabling enterprise-scale AI adoption by reducing costs, improving accuracy, and increasing speed. He argues that bespoke, narrowly focused models tailored to specific use cases are essential for delivering reliable, cost-effective AI solutions at scale. Learn more from The New Stack about the latest around enterprise development: Why Pure AI Coding Won’t Work for Enterprise Software How To Use Vibe Coding Safely in the Enterprise Join our community of newsletter subscribers to stay on top of the news and at the top of your game. 

AI coding assistants are boosting developer productivity, but most enterprises aren’t shipping software any faster. GitLab CEO Bill Staples says the reason is simple: coding was never the main bottleneck. After speaking with more than 60 customers, Staples found that developers spend only 10–20% of their time writing code. The remaining 80–90% is consumed by reviews, CI/CD pipelines, security scans, compliance checks, and deployment—areas that remain largely unautomated. Faster code generation only worsens downstream queues.GitLab’s response is its newly GA’ed Duo Agent Platform, designed to automate the full software development lifecycle. The platform introduces “agent flows,” multi-step orchestrations that can take work from issue creation through merge requests, testing, and validation. Staples argues that context is the key differentiator. Unlike standalone coding tools that only see local code, GitLab’s all-in-one platform gives agents access to issues, epics, pipeline history, security data, and more through a unified knowledge graph.Staples believes this platform approach, rather than fragmented point solutions, is what will finally unlock enterprise software delivery at scale. Learn more from The New Stack about the latest around GitLab and AI: GitLab Launches Its AI Agent Platform in Public BetaGitLab’s Field CTO Predicts: When DevSecOps Meets AIJoin our community of newsletter subscribers to stay on top of the news and at the top of your game.

Sean O’Dell of Dynatrace argues that enterprises are unprepared for a major shift brought on by AI: the rise of the developer. Speaking at Dynatrace Perform in Las Vegas, O’Dell explains that AI-assisted and “vibe” coding are collapsing traditional boundaries in software development. Developers, once insulated from production by layers of operations and governance, are now regaining end-to-end ownership of the entire software lifecycle — from development and testing to deployment and security. This shift challenges long-standing enterprise structures built around separation of duties and risk mitigation. At the same time, the definition of “developer” is expanding. With AI lowering technical barriers, software creation is becoming more about creative intent than mastery of specialized tools, opening the door to nontraditional developers. Experimentation is also moving into production environments, a change that would have seemed reckless just 18 months ago. According to O’Dell, enterprises now understand AI well enough to experiment confidently, but many are not ready for the cultural, operational, and security implications of developers — broadly defined — taking full control again.Learn more from The New Stack about the latest around enterprise developers and AI: Retool’s New AI-Powered App Builder Lets Non-Developers Build Enterprise AppsSolving 3 Enterprise AI Problems Developers FaceEnterprise Platform Teams Are Stuck in Day 2 HellJoin our community of newsletter subscribers to stay on top of the news and at the top of your game. 

In the era of agentic AI, attention has largely focused on data itself, while metadata has remained a neglected concern. Junping (JP) Du, founder and CEO of Datastrato, argues that this must change as AI fundamentally alters how data and metadata are consumed, governed, and understood. To address this gap, Datastrato created Apache Gravitino, an open source, high-performance, geo-distributed, federated metadata lake designed to act as a neutral control plane for metadata and governance across multi-modal, multi-engine AI workloads. Gravitino achieved major milestones in 2025, including graduation as an Apache Top Level Project, a stable 1.1.0 release, and membership in the new Agentic AI Foundation. Du describes Gravitino as a “catalog of catalogs” that unifies metadata across engines like Spark, Trino, Ray, and PyTorch, eliminating silos and inconsistencies. Built to support both structured and unstructured data, Gravitino enables secure, consistent, and AI-friendly data access across clouds and regions, helping enterprises manage governance, access control, and scalability in increasingly complex AI environments.Learn more from The New Stack about how the latest data and metadata are consumed, governed, and understood: Is Agentic Metadata the Next Infrastructure Layer?Why AI Loves Object StorageThe Real Bottleneck in Enterprise AI Isn’t the Model, It’s ContextJoin our community of newsletter subscribers to stay on top of the news and at the top of your game. 

Chris Aniszczyk, co-founder and CTO of the Cloud Native Computing Foundation (CNCF), argues that AI agents resemble microservices at a surface level, though they differ in how they are scaled and managed. In an interview ahead of KubeCon/CloudNativeCon Europe, he emphasized that being “AI native” requires being cloud native by default. Cloud-native technologies such as containers, microservices, Kubernetes, gRPC, Prometheus, and OpenTelemetry provide the scalability, resilience, and observability needed to support AI systems at scale. Aniszczyk noted that major AI platforms like ChatGPT and Claude already rely on Kubernetes and other CNCF projects.To address growing complexity in running generative and agentic AI workloads, the CNCF has launched efforts to extend its conformance programs to AI. New requirements—such as dynamic resource allocation for GPUs and TPUs and specialized networking for inference workloads—are being handled inconsistently across the industry. CNCF aims to establish a baseline of compatibility to ensure vendor neutrality. Aniszczyk also highlighted CNCF incubation projects like Metal³ for bare-metal Kubernetes and OpenYurt for managing edge-based Kubernetes deployments. Learn more from The New Stack about CNCF and what to expect in 2026:Why the CNCF’s New Executive Director Is Obsessed With InferenceCNCF Dragonfly Speeds Container, Model Sharing with P2PJoin our community of newsletter subscribers to stay on top of the news and at the top of your game. 

API sprawl creates hidden security risks and missed revenue opportunities when organizations lose visibility into the APIs they build. According to IBM’s Neeraj Nargund, APIs power the core business processes enterprises want to scale, making automated discovery, observability, and governance essential—especially when thousands of APIs exist across teams and environments. Strong governance helps identify endpoints, remediate shadow APIs, and manage risk at scale. At the same time, enterprises increasingly want to monetize the data APIs generate, packaging insights into products and pricing and segmenting usage, a need amplified by the rise of AI.To address these challenges, Nargund highlights “smart APIs,” which are infused with AI to provide context awareness, event-driven behavior, and AI-assisted governance throughout the API lifecycle. These APIs help interpret and act on data, integrate with AI agents, and support real-time, streaming use cases.IBM’s latest API Connect release embeds AI across API management and is designed for hybrid and multi-cloud environments, offering centralized governance, observability, and control through a single hybrid control plane.Learn more from The New Stack about smart APIs: Redefining API Management for the AI-Driven Enterprise How To Accelerate Growth With AI-Powered Smart APIs Wrangle Account Sprawl With an AI Gateway Join our community of newsletter subscribers to stay on top of the news and at the top of your game. 

A CloudBees survey reveals that enterprise migration projects often fail to deliver promised modernization benefits. In 2024, 57% of enterprises spent over $1 million on migrations, with average overruns costing $315,000 per project. In The New Stack Makers podcast, CloudBees CEO Anuj Kapur describes this pattern as “the migration mirage,” where organizations chase modernization through costly migrations that push value further into the future. Findings from the CloudBees 2025 DevOps Migration Index show leaders routinely underestimate the longevity and resilience of existing systems. Kapur notes that applications often outlast CIOs, yet new leadership repeatedly mandates wholesale replacement. The report argues modernization has been mistakenly equated with migration, which diverts resources from customer value to replatforming efforts. Beyond financial strain, migration erodes developer morale by forcing engineers to rework functioning systems instead of building new solutions. CloudBees advocates meeting developers where they are, setting flexible guardrails rather than enforcing rigid platforms. Kapur believes this approach, combined with emerging code assistance tools, could spark a new renaissance in software development by 2026.Learn more from The New Stack about enterprise modernization: Why AI Alone Fails at Large-Scale Code ModernizationHow AI Can Speed up Modernization of Your Legacy IT SystemsJoin our community of newsletter subscribers to stay on top of the news and at the top of your game.