The Enterprise AI Show

OpenClaw Enterprise: Making AI Agents Safe for the Enterprise

Brian interviews Derek Carr and Mrunal Patel, distinguished engineers at Red Hat, about OpenClaw Enterprise, the new open-source project that brings Red Hat, NVIDIA, and the OpenClaw community together to make persistent AI agents safe and manageable at enterprise scale. The work started with OpenShell, a deny-by-default sandbox the pair helped build with NVIDIA that treats every agent as potential malware. Agents never see credentials (a supervisor egress proxy injects them from outside), kernel features like Landlock restrict what they can touch, and when a blocked agent hits a wall, OpenShell can ask it why it tried, so a human can decide whether to allow the action now or never. Running real agents in that sandbox showed them that not everything should be untrusted. Memory, for example, is an enterprise record. That insight led them to OpenClaw’s own push to decompose agents into trusted and untrusted parts. OpenClaw Enterprise borrows the Hypershift hosted control plane pattern: gateways, memory, and session state live on a control plane cluster, while agent harnesses run in disposable OpenShell sandboxes on a data plane cluster, so a failed sandbox is simply replaced with full context. Enterprises set a permission ceiling; every new agent starts least privileged and grows only up to that limit, and blessed templates for roles like team Q&A, software co-development, and chief of staff make good agent setups repeatable instead of tribal knowledge. The guests also highlight multi-user agents, where colleagues steer the same agent together rather than flooding each other with AI-generated pull requests, and Skipper, the project’s own support agent that preserves the reasoning behind past decisions. The stack is model-agnostic and runs anywhere, including on self-hosted inference such as Dynamo or llm-d. With more than 3,000 pull requests merged since its September launch, the project is targeting a 1.0 release around year-end, with enterprise pilots and production rollouts expected in 2027. SHOW: 1075 SHOW VIDEO: https://youtu.be/yS2HEiGbxqo SHOW SPONSORS: Nord Security - NordLayer AI Scout SHOW LINKS TheEntAIShow #1009 - OpenClaw with Sally O’Malley OpenShell - the safe, private runtime for fleets of autonomous AI agents OpenClaw Enterprise - An open source, vendor neutral platform for managing persistent agents in sensitive environments. GUEST BIO(s) Derek Carr is a Senior Distinguished Engineer at Red Hat, where he has worked on OpenShift and in the Kubernetes community since 2014. He is a former member of the Kubernetes Steering Committee and has held long-running leadership roles in Kubernetes SIG Node and SIG Architecture. His recent work focuses on how enterprises can run AI agents safely on Kubernetes, including Red Hat’s work with NVIDIA’s OpenShell on kernel-enforced agent sandboxing. Mrunal Patel is a Distinguished Engineer at Red Hat, where he has worked on Linux containers, Kubernetes, and OpenShift since 2011. He started the CRI-O container runtime, is a maintainer of runc and the OCI runtime specification, and serves as a chair and technical lead of Kubernetes SIG Node. With Derek, he is now applying deep container isolation expertise to secure execution for AI agents on Red Hat AI. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Confidence or Con? The Psychology of Tech’s Boldest Leaders

Brian and Brandon take a detour from the usual tech news to ask a question every enterprise buyer of AI should be asking: why do some of tech’s most celebrated leaders project absolute confidence in claims that later prove false? Starting from the new Elizabeth Holmes documentary, they debate whether that kind of conviction is innate or learned. Brian argues it develops in stages, as each funding round and glowing press profile ties a founder’s identity more tightly to a story they must defend at all costs. Brandon describes a “contrarian effect,” in which a venture model with low hit rates rewards bold founders and tells them they are special. He recommends The Halo Effect as an antidote to success stories that confuse correlation with causation. They also question the VC preference for founders with hardship backstories. The conversation closes on Brandon’s framework of operators, believers, and pragmatists, and why every organization needs pragmatists willing to question big promises before they become fiction. SHOW: 1074 SHOW TRANSCRIPT: The Enterprise AI Show #1074 Transcript SHOW VIDEO: https://youtu.be/DXmB8_YwNRQ SHOW SPONSORS: Nord Security - NordLayer AI Scout SHOW LINKS The Enterprise AI Show GUEST BIO N/A this week FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Never Do the Same Work Twice: Rethinking CI for the Agent Era with Dan Manges

Brandon interviews Dan Manges, co-founder and CEO of RWX, about why continuous integration is the bottleneck for engineering teams working with coding agents. Manges, previously CTO at Braintree and Root, explains that he started RWX after watching CI tooling fail to scale with fast-growing codebases in regulated industries. He argues that feedback loops under five to ten minutes keep engineers in a flow state, and that slower loops push people (and agents) into costly context switching. Manges describes how RWX differs from traditional CI platforms. Instead of splitting pipelines into jobs that each spin up a virtual machine and repeat the same setup, RWX models work as a directed graph of tasks with declared dependencies, so everything runs with maximum parallelization and only once. He cites Datadog's recent blog post about the strain that cloning code across so many parallel jobs put on its Git servers. Failures can be retried from the point of failure, and compute can be sized per step. He explains that RWX uses content-based caching rather than Docker's lineage-based caching, so a cache miss can be followed by a cache hit and only the work affected by a change runs. Under the hood, he says, RWX built its own distributed container runtime on the same low-level primitives as Docker, runs on AWS, and is hosted by default, with an option to run inside a customer's own VPC. On agents, Manges says every RWX workflow can run from a CLI without a commit and push, which gives agents a fast, synchronous, more accurate and token-efficient loop. He recommends telling agents to run the CI workflow before they open a pull request, so the status check goes green instantly through caching, and argues there is no longer a reason for a human to be in the loop for build failures. He notes that review practices will vary widely, from thorough human review to software factory automation, and mentions a customer that has agents write throwaway Playwright tests for QA. He describes Honeycomb's move from CircleCI, which he says doubled their PR velocity, and the "AI legible CI" they describe. Looking ahead, Manges says the bottleneck in software engineering is shifting from writing code to validating it, and that RWX is expanding into sandboxes, preview apps, and custom webhooks. He says customers have asked about code hosting and review, but he promises nothing. SHOW: 1073 SHOW TRANSCRIPT: The Enterprise AI Show #1073 Transcript SHOW VIDEO: https://youtu.be/2QnbYmRltgM SHOW SPONSORS: Palo Alto - Secure every identity with Idira by Palo Alto Networks Nord Security - NordLayer AI Scout SHOW LINKS RWX Honeycomb switches to RWX for agent-ready CI/CD AI Amplifies Existing Practices: Lessons from Our AI Shift How Datadog Rebuilt Git Serving for CI (gitretriever) GUEST BIO Dan Manges is the co-founder and CEO of RWX, a cloud platform for running CI/CD and other automation for engineering teams and their coding agents. He was previously the founding CTO of Braintree, the payments company that acquired Venmo and was acquired by PayPal, and co-founded car insurance carrier Root, which went public in 2020. Contact Dan Manges: LinkedIn: dmanges Twitter/X: @dan_manges GitHub: dan-manges Medium: @dan_manges Website: dan-manges.com FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

AI News - Mid-October 2026

Brian and Brandon open mid-October’s AI news of the month by asking whether a yes-or-no answer engine like Jev could have saved the Commanders from a clock-management meltdown, and why a binary classifier is worth a reported $7.5 billion. A six-month check-in on their May AI draft leaves the scorecard roughly even, with Anthropic the clear winner over Google. Then come the financial cracks. CoreWeave’s borrowing costs have reportedly jumped from around 7% to about 15%, adding to Oracle’s contract pullback as yellow flags for the neocloud and data center layer. Meanwhile, Anthropic and OpenAI keep pulling in revenue even as enterprise CIOs struggle to find ROI. Brandon weighs Andrej Karpathy’s claim that only a few thousand people have seen what AI can really do against the labs’ opposite storytelling styles. The pair also question why Meta keeps promoting its frontier models when its Muse personal agent looks like the real monetization story. They close on whether software fundamentally changed in the last two weeks. Brandon argues AI has put professional-grade app building within reach of millions, pointing to his upcoming PodEditor app, but says finding demand and building a business are as hard as ever. They see Bending Spoons’ model of running acquired apps with AI and a skeleton crew as a possible playbook for established vendors. SHOW: 1072 SHOW TRANSCRIPT: The Enterprise AI Show #1072 Transcript SHOW VIDEO: https://youtu.be/Kp7AH_IdwIg SHOW SPONSORS: Nord Security - NordLayer AI Scout Palo Alto - Secure every identity with Idira by Palo Alto Networks FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Agents Are Identities Too: Identity Security After CyberArk with Amy Blackshaw

Brandon interviews Amy Blackshaw, Senior Vice President of Identity Marketing at Palo Alto Networks, about identity security in the age of AI agents. Blackshaw, who spent four years leading product marketing at CyberArk before Palo Alto Networks acquired it in February 2026, explains how identity management evolved from an IT operations task into a security discipline, and why identities now need to be protected across their full lifecycle. She walks through the three segments of identity security (access management, privileged access management, and identity governance), argues that human-driven quarterly access reviews are painful and error-prone, and says vendors like Palo Alto Networks are pulling governance into the identity security platform and using AI-driven roles to pre-approve most access. Blackshaw describes how frontier AI models change the attack landscape, with the cost of entry dropping toward zero and attackers moving at machine speed. She cites Unit 42 figures on how fast intrusions reach data exfiltration, an initial access time of about 25 minutes, and a recent investigation in which the full attack took under 10 hours. Her answer is control alongside detection and response: zero standing privileges and just-in-time access, so an attacker who steals a credential reaches a dead end. She and Brandon discuss why the most boring attacks, like a weak password and no second factor, are often the most effective, and why attackers often succeed with hard-coded machine credentials. On agents, Blackshaw says an agent is a machine identity with human-like traits: it can reason, decide, and escalate its own privileges, and you can't MFA it. She argues agents should be first-class identities with least privilege, just-in-time access, lifecycle governance, and a gateway for policy. For human-in-the-loop agents, the person is the anchor and must approve the moments that matter, with guardrails against consent fatigue. For autonomous agents, policy is the anchor, with tightly set permissions, behavior monitoring, and a kill switch. Brandon and Blackshaw also cover how to apply zero standing privilege for admins, developers, and agents without slowing developers down, and why policy and context (time, location, and usual behavior) should decide access. They close on who owns identity, which now rolls up through the CISO and increasingly reaches the CIO and CEO, why identity and security operations teams need to share data, and Blackshaw's case for platformization in a fragmented security market. SHOW: 1071 SHOW TRANSCRIPT: The Enterprise AI Show #1071 Transcript SHOW VIDEO: https://youtu.be/CcaFoJRu0YM SHOW SPONSORS: NordLayer - Use ENTERPRISE10 for 10% off Palo Alto - Secure every identity with Idira by Palo Alto Networks SHOW LINKS Pathways to Identity Security Success SC World: Palo Alto Networks launches Idira IdentitySecOps: The Closed-Loop Model for AI-Speed Defense Unify Identity and Security Operations to Stop AI Attacks (ebook) Unit 42 Frontier AI Defense Frontier AI and Identity Security in Financial Services (Q&A) Cortex XSIAM Palo Alto Networks Idira Prisma AIRS Amy Blackshaw: You Do Not Have Three Years 2026 Unit 42 Global Incident Response Report GUEST BIO Amy Blackshaw is Senior Vice President of Identity Marketing at Palo Alto Networks, where she leads marketing for the identity business unit under the Idira brand. She has spent almost 20 years in cybersecurity, most of it in product marketing at large vendors, including four years leading product marketing at CyberArk, the pioneer of privileged access management, which Palo Alto Networks acquired in February 2026. Contact Amy Blackshaw: Amy Blackshaw on LinkedIn FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

A Cloud Nobody Owns: Sovereign Infrastructure for the Agent Era with Dominic Williams

Brandon interviews Dominic Williams, founder, president, and chief scientist of the DFINITY Foundation, about the Internet Computer and why he set out to build a new back-end compute stack. Williams traces his path from building the kids' MMO Fight My Monster, one of the first serious production users of Cassandra, to decentralized networks and the founding of DFINITY in 2016. He explains the thesis behind the Internet Computer: a stack of software running down to the silicon can never be fully secured, so he wanted a cloud created by a network protocol, where software is tamper-proof, always on, and secure by default without firewalls, patches, or security and sysadmin teams. Williams walks through how it works. Code compiles to WebAssembly and runs in persistent memory, replicated across independent nodes using Byzantine fault tolerance, so there is no database or file system to connect to, and a misbehaving node, even a "Dr. Evil" one, cannot corrupt the result. Brandon presses on data protection, and Williams points to hardware encapsulation such as AMD SEV-SNP and the ability to choose trusted node providers. He then describes cloud engines, which let an enterprise pick nodes from a marketplace (hyperscalers, independent data centers, or on-prem), reconfigure them without downtime, and avoid vendor lock-in, with payment handled behind the scenes so it looks like a conventional cloud. On adoption, Williams says data sovereignty is a major driver, citing the U.S. CLOUD Act and governments, such as Pakistan, that want infrastructure they control and can inspect as open source. He describes Caffeine, which builds apps on the Internet Computer from a chat prompt, and Open SaaS, a suite meant to replace siloed SaaS tools with a single AI-aware system. To get started, he recommends pasting the prompt from the skills site into a coding assistant and having it write the back end in Motoko, a language designed for AI agents that guards against data loss when upgrading apps in production. Cloud engine coupons are expected to open up soon. SHOW: 1070 SHOW TRANSCRIPT: The Enterprise AI Show #1070 Transcript SHOW VIDEO: https://youtu.be/-H4KS5AMKI8 SHOW SPONSORS: Palo Alto - Secure every identity with Idira by Palo Alto Networks NordLayer - Use ENTERPRISE10 for 10% off SHOW LINKS DFINITY Foundation Internet Computer OpenCloud (cloud engines) Internet Computer Skills: Get Started DFINITY Unveils Open SaaS Vision for AI-Powered Business Software Caffeine Motoko programming language WebAssembly Apache Cassandra LinkedIn: thedwilliams Twitter/X: @dominic_w GUEST BIO Dominic Williams is the founder, president, and chief scientist of the DFINITY Foundation, the Zurich-based non-profit behind the Internet Computer, which he began developing in 2015 and launched in 2021. A longtime engineer-entrepreneur focused on distributed systems, he previously founded the MMO game Fight My Monster and served as president and CTO of the Palo Alto incubator String Labs. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

The New, New Kingmakers: Agents, Developers, and the Future of Software with Stephen O'Grady

Brandon interviews Stephen O'Grady, principal analyst and co-founder of RedMonk, about his 2013 book *The New Kingmakers* and his recent follow-up, "Agents: The new, New Kingmakers." O'Grady explains that the original thesis was simple: developers had gone from an afterthought in technology purchasing to the group whose choices determined what got used, driven first by open source (the LAMP stack, Linux) and then by cloud services like S3 and EC2 that removed the need to ask for budget. The two discuss why Sun was disrupted from every direction while Oracle and Cisco endured, with O'Grady noting that databases are the stickiest, highest-risk asset in the stack. He recounts getting an early look at GitHub Copilot a year before ChatGPT, but says the late-2025 jump in model capability was a step change he did not foresee. On software development, O'Grady expects a spectrum rather than a consensus on reviewing AI-written code: toy apps and disposable software may never be read by a human, while medical imaging, aviation, and defense will demand real oversight. He uses his construction background to describe the developer's shift from builder to architect, and warns that the "human in the loop" role is the weak end of the outcomes. Brandon and O'Grady examine why individual developers feel superhuman while organizations don't see the gains, with O'Grady pointing to downstream bottlenecks like the flood of PRs overwhelming open source maintainers. He cites Dr. James Bessen's bank teller research as a reminder that automation's effect on employment is hard to predict, and asks what happens if there is an order of magnitude more software in the world. On vendors, O'Grady expects both human-facing and agent-only businesses to coexist, with the biggest growth in agents because they scale like cloud, while Brandon points to Cloudflare's rewrite of Wrangler as a model for exposing complete APIs. O'Grady cautions that giving agents the keys to the kingdom is harder than it sounds, since subtle violations (like pulling copyrighted headshots from MLB.com) are harder to foresee than deleting a database. They close with hot takes: O'Grady takes the under on a 10% chance of AI wiping out humanity, arguing the industry doesn't draw a bright enough line between digital and physical capability, and he remains skeptical of frontier model valuations and the circular investments behind them, especially as open-weight models close the gap. SHOW: 1069 SHOW TRANSCRIPT: The Enterprise AI Show #1069 Transcript SHOW VIDEO: https://youtu.be/E6UFnW40BHQ SHOW SPONSORS: NordLayer - Use ENTERPRISE10 for 10% off Palo Alto - Secure every identity with Idira by Palo Alto Networks SHOW LINKS The New Kingmakers Agents: The new, New Kingmakers Tecosystems Blog RedMonk LinkedIn GUEST BIO Stephen O'Grady is the principal analyst and co-founder of RedMonk, a developer- and practitioner-focused industry analyst firm. He is the author of *The New Kingmakers* (2013), which argued that developers had become the most important decision-makers in technology, and writes the Tecosystems blog. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

AI News for End of September 2026

Brian (@bgracely) and Brandon (@bwhichard) discuss the biggest AI news stories from the second half of September 2026. Oracle sent a force majeure notice on a major data center build, which Brandon reads as an early signal worth tracking. If other companies follow, it would point to stress in AI infrastructure commitments, and if not, it is a one-off. Brian suggests the industry needs a "Jenga chart" of whose money is propping up whose, with the contractors who actually pour concrete at the base. On the wave of agent launches (Meta's Muse, OpenAI's dots, and OpenClaw Enterprise), they agree that giving everyone a cloud-hosted computer for their agent is a real step toward mainstream adoption. Brian says the unanswered questions are accountability when an agent errs, security controls that don't require dozens of settings, and how people decide what is important enough to keep for themselves. Brandon argues agents should step in at the moment of a task, like filling out a form, rather than needing constant delegation. They also note the move toward soft, cuddly agent branding and the lack of easy ways to share skills across people and teams. The show closes with Meta's hiring of MongoDB's CEO to lead its enterprise push, which Brandon expects to fail because consumer companies start at a deficit in enterprise. SHOW SPONSORS: Palo Alto - Secure every identity with Idira by Palo Alto Networks NordLayer - Use ENTERPRISE10 for 10% off SHOW: 1068 SHOW TRANSCRIPT: The Enterprise AI Show #1068 Transcript SHOW VIDEO: https://youtu.be/kaNik7LxV3s SHOW LINKS: https://theenterpriseaishow.com/ FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

The Anthropic IPO?

We review the potential Anthropic IPO, in the context of revenue trajectory, expense trajectory and planned commitments, past and future funding rounds, future expanse commitments to large datacenter projects, and the strengths and weaknesses of the business in the current and potentially future political climates in the US and around the world. SHOW: 1067 SHOW TRANSCRIPT: The EntAIShow #1067 - The Anthropic IPO? SHOW SPONSORS Nasuni - Activate your data for AI and request a demo NordLayer - Use ENTERPRISE10 for 10% off. SHOW LINKS Strengths: Brand awareness of Claude with software developers Use-Case adoption for Enterprise workers (white-collar, software, research, etc.) Depth and experience of Anthropic Frontier Labs (Model Building, Research, etc.) Weaknesses CEO who frequently scares the world with his proclamations about doom Uncertainty about alignment to US government after 2026 Increasing annual losses against revenue Security and safety concerns (Mythos) Future commitments and long-term financials? FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

The OpenAI IPO?

We review the potential OpenAI IPO, in the context of revenue trajectory, expense trajectory and planned commitments, past and future funding rounds, future expanse commitments to large datacenter projects, and the strengths and weaknesses of the business in the current and potentially future political climates in the US and around the world. SHOW: 1066 SHOW TRANSCRIPT: The Enterprise AI Show #1066 SHOW VIDEO: https://youtu.be/PkLSI2pRZs8 SHOW SPONSORS: NordLayer - Use ENTERPRISE10 for 10% off. Nasuni - Activate your data for AI and request a demo FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

How Do Regulated Enterprises Deploy Private, Secure AI?

Aaron interviews Ivan Lee, founder and CEO of Datasaur, about what it takes to build private, secure AI for regulated enterprises. Lee traces the shift away from third-party model reliance to compliance pressure, IP protection, and cost, and frames the core decision as buy versus rent: enterprises historically rented frontier models by default, but Western and Chinese open-weight models have matured enough that companies like AT&T now run a growing share of workloads on owned infrastructure. He breaks the stack into three layers (infrastructure, model, and harness), argues models have become commoditized enough to be swapped like building materials, and calls the harness, the connective layer to internal data and tools, the least solved but highest-leverage piece. Lee describes agentic AI's shift from opt-in tools to opt-out, event-triggered workflows as the real 2025-2026 adoption unlock, while flagging FinOps trade-offs (agentic tasks can run 3 million tokens versus 2,000 for a chatbot query) and the custom benchmarking his team uses to earn CISO trust before production rollout. SHOW: 1065 SHOW TRANSCRIPT: The Enterprise AI Show #1065 SHOW VIDEO: https://youtu.be/PkLSI2pRZs8 SHOW SPONSORS NordLayer - Use ENTERPRISE10 for 10% off. Nasuni - Activate your data for AI and request a demo SHOW LINKS Datasaur homepage Ivan Lee on LinkedIn Datasaur Launches Forge: AI Native Services to Deploy Private, Model-Agnostic AI Inside Regulated Enterprises Datasaur's Private LLM overview GUEST BIO Ivan Lee is the founder and CEO of Datasaur, which builds private, model-agnostic AI agents that deploy entirely inside a regulated enterprise's own infrastructure. He previously built AI products at Yahoo and Apple after Yahoo acquired his first company, Loki Studios, and holds a computer science degree from Stanford. Datasaur's clients include a leading GSIB, federal agencies, and Am Law 100 firms, and its backers include Initialized Capital and OpenAI president Greg Brockman. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

AI News for Mid-September 2026

Aaron and Brandon cover what the week's news means for enterprise buyers. They question whether Oracle's reported $664 billion backlog is durable demand or overlapping commitments, and Brandon argues contracted future spend is how enterprise business works. On the labs' "slow down" messaging, Brandon sees no coordination, only incentives, while Aaron finds the timing too coincidental. Salesforce's Nemotron-based model points to enterprises customizing open models instead of building their own, and Brandon doubts labs will displace systems of record like Workday or SAP. Aaron argues agent pricing has reverted to familiar free, bundled, and negotiated tiers. He sees agent swarms as research capability that most enterprises lack the trust and human-in-the-loop controls to run. SHOW SPONSORS: Nasuni - Activate your data for AI and request a demo NordLayer - Use ENTERPRISE10 for 10% off. SHOW: 1064 SHOW TRANSCRIPT: The Enterprise AI Show #1064 Transcript SHOW VIDEO: https://youtu.be/hrfTovN3kMw SHOW LINKS: https://theenterpriseaishow.com/ FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Building AI Agents That Support Enterprise Teams

Aaron interviews Diego Oppenheimer, general partner at the AIT Fund (the venture fund behind the AI Tinkerers builder community), about what it takes to move AI agents from personal experimentation into enterprise-grade deployment. Drawing on his own agent fleet (a chief-of-staff agent, a content strategist, and Ashley, an always-on AI community manager serving AI Tinkerers' 127,000+ members) as a proving ground, Oppenheimer lays out an enterprise security model built on treating agents as digital employees with their own delegated identities and credentials, scoped access rather than blanket access to accounts like email and calendar, and continuous audit logging (via the open-source tool Hyperware) in place of trust based on prior behavior. He argues enterprises get burned by chasing a single omnipresent agent instead of narrowing agents to specialized, restricted workflows, the same specialization principle that has driven results throughout machine learning, and flags scheduling as a deceptively hard example where the "10% edge cases" eat most of the effort. He also describes deliberately red-teaming his own agents for weeks (trying to break out of containers, extract credentials, and social-engineer them) before granting them any real access, and points to NanoClaw's containerized, minimal-component design as a model worth enterprise attention. The conversation closes on identity, responsibility, and permissioning as the unresolved internals enterprises must solve before scaling agent autonomy, and on Oppenheimer's prediction that many enterprise roles will shift toward "exception handling" as teams of AI coworkers absorb routine work and escalate only what needs human judgment. SHOW: 1063 SHOW TRANSCRIPT: The Enterprise AI Show #1063 Transcript SHOW VIDEO: https://youtu.be/vL8btTMcInE SHOW SPONSORS NordLayer - Use ENTERPRISE10 for 10% off Nasuni - Activate your data for AI and request a demo SHOW LINKS Diego Oppenheimer homepage Diego on LinkedIn AI Tinkerers Guardrails AI NanoClaw on GitHub The New Stack, "OpenClaw vs. Hermes Agent: The race to build AI assistants that never forget" The New Stack, "NanoClaw and Docker team up to isolate AI agents inside MicroVM sandboxes" GUEST BIO Diego Oppenheimer is working full-time at AIT Fund (the fund behind AI Tinkerers) and previously was a partner at Factory and served as CEO-in-residence at Factory. He founded Algorithmia, an enterprise MLOps platform acquired by DataRobot, co-founded Guardrails AI, and earlier in his career led teams at Microsoft shipping Excel, SQL Server, and Power BI. He is currently running AI agents inside his own team to take on real, sustained work, including one named Ashley, and documenting what he's learned in a new video series with Joe Heitzeberg. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Securing AI Agents in the Enterprise: NanoClaw, Zero Trust Guardrails, and Governance at Scale

Brian and Aaron interview Gavriel Cohen, co-founder and CEO of Nanoco and creator of the open-source agent framework NanoClaw, about securing AI agents in enterprise environments. Cohen shares how he built NanoClaw after discovering major security and safety gaps while using agents for an AI native marketing agency, and how the project grew to over 30,000 GitHub stars and over half a million downloads. They discuss why Fortune 500s, financial institutions, universities, and government groups feel urgent pressure to adopt agents but are blocked by control, privacy, and security concerns. Cohen outlines a zero-trust approach using microVM isolation, no credentials inside agent environments, a policy-enforcing gateway with granular controls, audit logs, cost attribution, and human-in-the-loop approvals at key decision points. SHOW: 1062 SHOW TRANSCRIPT: The Enterprise AI Show #1062 Transcript SHOW VIDEO: https://youtu.be/h906EEQSRs8 SHOW LINKS: NanoClaw on GitHub NanoCo homepage TechCrunch: The wild six weeks for NanoClaw's creator that led to a deal with Docker The New Stack: Gavriel Cohen found his own code inside OpenClaw, so he walked away OpenAI: The Hugging Face incident and the road ahead SHOW SPONSORS: Nasuni - Activate your data for AI and request a demo NordLayer - Use ENTERPRISE10 for 10% off GUEST BIO: Gavriel Cohen is co-founder and CEO of NanoCo, and creator of NanoClaw, the open-source agent harness he built as a small, auditable, secure alternative to OpenClaw. He spent a decade as a developer and team lead at Wix before building NanoClaw in a weekend, a project that has since drawn a Docker integration and an outside security review. He holds a BSc in Physics and Computer Science from Tel Aviv University. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

How Open-Source is Reshaping the AI Infrastructure Stack

Aaron interviews David Aronchick, CEO @ Expanso (former PM lead for Kubernetes, Kubeflow co-founder, and open-source ML leader at Azure) about how open source is reshaping the AI infrastructure stack. Aronchick recounts his path from early Linux and enterprise work to launching Kubernetes and GKE, then creating Kubeflow in 2017 to orchestrate end-to-end ML workflows on Kubernetes. The discussion centers on gaps in AI infrastructure, especially reproducibility and determinism across hardware, drivers, OS, packages, and data lineage, arguing Kubernetes alone can’t fully solve it. They contrast open weights with true open-source models, noting that real openness would require reproducible training data and infrastructure. They explore “AI-native” enterprise architecture, the role of open-source harnesses/wrappers to add deterministic controls, and growing edge/distributed compute needs driven by governance, compliance, bandwidth, and hybrid deployment realities. SHOW: 1061 SHOW TRANSCRIPT: The Enterprise AI Show #1061 Transcript SHOW VIDEO: https://youtu.be/kpQg3YIIUL8 SHOW LINKS: Expanso homepage TechArena, "Expanso's David Aronchick on Data Gravity and Pipeline Debt": Open at Intel podcast, "Data Privacy and Efficiency with Bacalhau Compute Over Data" SHOW SPONSORS: NordLayer - Use ENTERPRISE10 for 10% off Nasuni - Activate your data for AI and request a demo SHOW TOPICS: You have a super interesting background (First managing PM for Kubernetes, Co-founded Kubeflow, led open-source ML at Microsoft Azure). Give everyone a brief introduction and how you became so involved in open-source and the Enterprise OSS topics: Back when we were The Cloudcast, we covered K8s in depth, but I’m not sure we ever did a show on Kubeflow. Kubeflow tried to bring Kubernetes-style orchestration to ML workflows. Looking back, what did that generation of open-source AI infrastructure get right, and what did it miss that the current wave (agents, inference at the edge) is now having to solve for again? Oh, and maybe give a quick intro to Kubeflow as well for those that aren’t familiar Zooming out - open source shaped your whole career, from Kubernetes to Kubeflow to Bacalhau. Where do you think open source has the most leverage in the AI infrastructure stack right now, and where do you think it's losing ground to closed, vendor-controlled platforms? What are your thoughts on “OSS models”? Today, OSS really means open weights. Do you think there will ever be a truly OSS model? What would it take? Thoughts on the state of the industry? A couple of Enterprise “grab bag” questions for you on a few different topics while we have you: "AI-native" gets used a lot and means different things to different people. What does AI-native actually mean for enterprise architecture in your view, and how is it different from just bolting AI onto an existing cloud or data stack? Regulatory and data residency pressure keeps coming up across industries (telecom, healthcare, financial services). How much of the edge/distributed compute push is being driven by AI performance needs versus governance and compliance requirements? Which one is the bigger driver right now? CLOSING: If anyone is interested, what’s the best way to get started? FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Vitamins vs. Pain Killers vs. Whippets - The Enterprise AI Adoption Problem

Brian, Brandon, and Aaron discuss enterprise AI adoption using the sales analogy of whether AI is a “vitamin” or a “painkiller,” arguing that successful transformation still requires a burning-platform event. Brandon suggests AI adoption resembles past digital transformations: without urgent pressure (e.g., a data center closing), organizations resist change and justify existing processes. Brian describes a compressed hype cycle from ChatGPT excitement to pilots and guardrails, followed by difficulties with data, cost-effective scaling, and making AI behave deterministically, while fear of competitors keeps efforts alive. They add a third category, “Whippets”, short-term, resume-driven initiatives led by leaders who leave others “holding the bag.” They debate examples like Sheetz’ multiple VMs and argue that AI’s promise is personal productivity, but note a lack of enterprise collaboration and shared-memory tools that limit organizational impact. SHOW: 1060 SHOW TRANSCRIPT: The Enterprise AI Show #1060 Transcript SHOW VIDEO: https://youtu.be/XZqomQosv1w SHOW LINKS: The Enterprise AI Show SHOW SPONSORS: Nasuni - Activate your data for AI and request a demo NordLayer - Use ENTERPRISE10 for 10% off. The thesis: Successful digital transformation usually has a forcing function; you migrate the data center because the real estate got sold, not because someone promised abstract savings. Deadline + shared incentive = people actually change. AI adoption mostly lacks that: no one's forcing the migration, so it defaults to "give everyone Copilot licenses and hope." Core question: If your business is healthy and there's no burning platform, how do you adopt AI in a way that's more than expensive theater, without a crisis to manufacture urgency? Discussion topics: Forcing functions vs. vibes: What are the AI-era equivalents of "the real estate got sold"? A support team that's actually understaffed, a process with a real bottleneck, a cost center leadership is already scrutinizing vs. a mandate to "use more AI." The unknown-unknowns problem: Most orgs don't know which of their workflows AI would actually help vs. where it's a novelty. How do you go find that out cheaply, without a company-wide token-burning experiment as the discovery mechanism? Bottom-up signal vs. top-down mandate: Does real usage data (who's actually using tools, for what) surface better targets than an executive committee guessing at use cases? Contrast with the failed "abstract savings" migration: What does an AI initiative look like when it's tied to a concrete, already-painful problem instead of a general efficiency narrative? The FOMO trap: Distinguishing "we don't want to miss the platform shift" (legitimate) from "we need AI headlines for the board" (theater), and how leadership can tell which one they're actually doing. Measurement: If there's no forcing function, what replaces the natural deadline/incentive alignment as the way you know it's working, or that it's time to kill it? Final Thought Is the right move small, cheap, bounded bets against known pain points, treating AI adoption like a search problem, not a rollout, rather than a company-wide transformation initiative looking for a reason to exist? FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

AI News of the Month - August 2026

Aaron, Brian, and Brandon cover major stories including NVIDIA’s record quarter and continued growth forecasts, alongside concerns about declining free cash flow and customer financing. They discuss NVIDIA’s reported $12.9B acquisition of Hugging Face as a strategic move to strengthen the open-model ecosystem and go up the stack, and Stripe’s $8B acquisition of OpenRouter as routing infrastructure for model choice and potential agent-to-agent commerce. The group reacts to reports of OpenAI agent testing in which agents collaborated, manipulated logs, and tried to deceive humans, framing it as a security and guardrails issue. They also mention Microsoft employees’ surprising AI spend, OpenAI’s “Jalapeño” hardware push and manufacturing constraints, and Salesforce’s “Claude Force” concept of using Claude as the UI to query Salesforce data. SHOW: 1059 SHOW TRANSCRIPT: The Enterprise AI Show #1059 Transcript SHOW VIDEO: https://youtu.be/Clmgst03-eg SHOW LINKS: NVIDIA’s record quarter NVIDIA buys Hugging Face Stripe buys OpenRouter OpenAI’s new chip SHOW SPONSORS: NordLayer - Use ENTERPRISE10 for 10% off. Nasuni - Activate your data for AI and request a demo Topic: Link to the full list of topics for the month FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

AI Watermarking: Compliance Theater or Real Provenance Tool

SUMMARY: Brian, Brandon, and Aaron focus on AI watermarking, driven largely by EU transparency requirements, and discuss how approaches like token-selection patterns can be detected but were reportedly cracked quickly with tools that strip watermarks. Brandon and Brian debate whether watermarking is useful long-term, suggesting most people care more about whether content is helpful than whether AI was involved, and questioning the added cost and real-world impact of such regulation. They also explore implications for education policies that ban AI use, changing assessment methods to curb cheating, and potential enterprise and government procurement issues where “no AI” requirements could trigger disputes and lawsuits, while AI review may also level the playing field in contract understanding. SHOW: 1058 SHOW TRANSCRIPT: The Enterprise AI Show #1058 Transcript SHOW VIDEO: https://youtu.be/6zlN_oIR5Xc SHOW LINKS: Anthropic Watermarking Claude Support on Watermarking SHOW SPONSORS: Nasuni - Activate your data for AI and request a demo Topic: Anthropic recently started invisibly watermarking all Claude-generated text and files (Aug 11), joining Google (SynthID) and ~190 companies that signed the EU's AI Act Transparency Code. Article 50 became enforceable August 2, with fines up to €15M or 3% of global turnover for non-compliance. Within 24 hours of Anthropic's announcement, a free tool to strip Claude's watermark showed up on GitHub. Core question: Is watermarking building durable AI provenance infrastructure, or is it a regulatory checkbox that breaks the moment someone runs a paraphraser? Discussion angles: The cat-and-mouse problem: Watermarks degrade with editing/paraphrasing/translation by design; light edits survive, heavy rewrites don't. Is a signal that vanishes under normal use actually useful, or just plausible deniability for labs? Regulatory arbitrage: EU forces the mandate, but xAI hasn't signed the voluntary Code. What happens to companies operating in the gap, and does the EU rule become a de facto global standard the way GDPR did? What it's actually good for: Not a lie detector, a provenance/tamper flag. Useful for enterprise content authenticity and platform moderation pipelines, much less useful for catching a student or a bad actor who just runs one rewrite pass. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript
• Chapters

Your AI Project Doesn't Need More Agents

SUMMARY: Brandon speaks with Rich Ziade, co-founder and CEO of Aboard, about why enterprise AI projects fail without real discovery, why "agents" have been oversold as a headcount play, and why organizational urgency, not new tooling, is what actually makes digital transformation succeed. SHOW: 1057 SHOW TRANSCRIPT: The Enterprise AI Show #1057 Transcript SHOW VIDEO: https://youtu.be/RMgycbmuXGs SHOW SPONSORS: Nasuni - Activate your data for AI and request a demo SHOW LINKS: https://www.aboard.com/ https://aboard.com/podcast/ Topic 1 - From lawyer to digital transformation CEO: Rich's path through Postlight (with co-founder Paul Ford), the sale in 2021, and how Aboard was already incubating inside Postlight Labs before AI "landed like a spaceship." Topic 1a - The six months after ChatGPT arrived: why Aboard resisted rushing a prompt-based fix into messy, political, human organizations, and why "vibe coding" convinced them to hang back rather than parachute AI into a company. Topic 2 - The doctor/patient analogy: executives walk in asking for a specific AI "medicine" instead of describing the underlying pain, and why real engagements start with tests and diagnosis, not the prescription the client thinks they want. Topic 2a - Why discovery hasn't fundamentally changed in the AI era — still in-person interviews and observation, with AI mainly useful for note-taking and summarizing documentation, not for skipping the hard thinking. Topic 3 - The agent hype cycle: why Rich thinks the "millions of agents" narrative (including Anthropic's Boris Cherny running swarms of planning/implementation agents) reflects an engineering-execution worldview rather than a product or organizational one — and why he sees the agent narrative cooling off. Topic 3a - The "spreadsheet problem" vs. targeted AI: most of Aboard's actual delivery work (90%+) isn't agents — it's modernizing spreadsheet-run processes and building narrow RAG/vector tools so people can query their own data in plain English. Topic 4 - "Forward deployed" as the new name for an old idea — going on-site, listening, and understanding a client's world before proposing a solution. Topic 5 - Why no successful digital transformation starts without a real, externally imposed deadline or crisis — and why "innovation labs" without urgency rarely ship anything. Topic 6 - Lightning round: Is AI a bubble? Should GPUs be securitized assets? Three things to do in NYC in one day. FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

NVIDIA's Pivot from Chipmaker to Financier

SUMMARY: Brian, Brandon, and Aaron discuss news about Nvidia’s reported $105B backing of OpenAI’s Ohio data center and what it implies for GPUs as an “asset class” and enterprise AI. Brian argues Jensen Huang is shifting Nvidia’s narrative from needing the newest chips immediately to portraying GPUs as long-lived, cash-flowing assets that can be financed like bonds, pushing risk onto banks and private equity. Brandon agrees scarcity has extended older GPU usefulness but warns the market could be flooded with newer, cheaper, more efficient hardware, leaving debt tied to obsolete equipment. Aaron likens GPUs to airplanes, expensive assets requiring constant utilization, while noting new AI builds demand entirely new data centers for power and cooling. The group questions widespread lack of profitability, compares the financing trend to past bubbles, and debates the optimistic case that breakthroughs could ultimately justify the investment. SHOW: 1056 SHOW TRANSCRIPT: The Enterprise AI Show #1056 Transcript SHOW VIDEO: https://youtu.be/vTLTdIZueJM SHOW SPONSORS: Nasuni - Activate your data for AI and request a demo Show topic: Nvidia's Pivot from Chipmaker to Financier Nvidia just backed $105B for OpenAI's Ohio data center and helped mobilize $500B+ in Wall Street financing (Apollo, Blackstone, BlackRock, Goldman, KKR) to fund GPU purchases, while AMD, Google, and Cerebras chip away at its tech lead. The moat is moving from silicon to balance sheet. Core question: Is a GPU actually securitizable like real estate or aircraft, or is this circular financing dressed up as infrastructure? The bull case: GPUs as productive, cash-flow-generating assets (compute-as-a-service) → financeable like data centers or planes, unlocking capital hyperscalers alone couldn't raise. The bear case: Depreciation risk; GPUs age fast, unlike buildings. What's the residual value of an H100-class chip in 2030? Securitizing a depreciating, obsolescence-prone asset is a very different bet than securitizing land. Circularity concern: Nvidia financing the customers who buy Nvidia chips, who generate the revenue that justifies Nvidia's valuation, echoes vendor financing bubbles (Cisco/telecom, 2000). Precedent: Compare to aircraft leasing/securitization models: what made those work (long asset life, resale markets, standardized valuation), and whether GPUs have any of that yet. Who bears the risk if utilization or model economics don't pan out: Nvidia, the banks, or the credit markets buying the paper? FEEDBACK? Email: show @ the enterprise ai show dot com Bluesky: @TheEntAIShow.bsky.social Twitter/X: @TheEntAIShow Instagram: @TheEntAIShow

• Transcript