Episode 22 - Secure Agent Authorization: Tasks as the Authorization Boundary for Enterprise Agents

#AgenticAI #AIAgents #AIProductivity #AIAutomation #CodingAgents #EnterpriseAI #SoftwareDevelopment #AgenticMesh Agent Harnesses have evolved radically from humble chatbot beginnings.  Now they help an agent create child agents, even swarms of them, that can be coordinated to complete larger projects.  But the direction of evolution is towards collaborating ecosystems of agents.  Today's Agent Harnesses are designed to be controlled by a single user, operating in a single environment, using the user's identity and permissions, using relatively simple local interprocess communications.But agent use cases and deployment patterns are changing.  Early adopters are building agent ecosystems, where agents are "headless", autonomous, and unbound from any direct user control.  They operate under their own identity and have permissions and authorization to execute tools based upon very strict controls.  And they are distributed computing entities in their own right, becoming services accessible via a ubiquitous Communications and Collaboration Fabric.So, today, the focus may still be on the individual agent and its Agent Harness, but once we have thousands, maybe even millions, of agents in each enterprise, then it is the Agent Ecosystem Harness that needs our attention.Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

#AgenticAI #AIAgents #AIProductivity #AIAutomation #CodingAgents #EnterpriseAI #SoftwareDevelopment #AgenticMesh Agents create great demos. They look cool, but when we really look at them, they are still pretty simple. Most times they address only a slice of the business need, let any user read every record, or leave no logs showing who changed what. Maybe they run on a laptop for a single use when the target system needs to run on a cloud cluster. Maybe they work for one user when the actual solution must serve hundreds.We think this is the 80/20 rule in action. When we look at the demo, we get an exaggerated impression that much of the work is done (the 80%) and that there is relatively little remaining (the 20%).  Clearly, these numbers describe how the demo feels, but our experience is that they tell us little about the effort remaining.☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture #AISovereignty #AIInfrastructure #AIDataThe problem is not really agent judgement.  Rather, it is because of the way we communicate with agents that leads to wrong assumptions, incomplete information, and, ultimately, poor results.☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips Chapters00:00 – Do AI Agents Really Lack Judgment?01:43 – What “Taste and Judgment” Actually Means03:04 – The Real Problem: Missing Context07:15 – Why Iteration Is Your Friend10:03 – Agile at Agent Speed14:48 – Building Shared Workspaces for Agents19:24 – How Agent Teams Can Work Together20:47 – The Challenge for Knowledge Worker Agents23:13 – Giving Agents the Right Context25:37 – Privacy, Security & Specialized Agents26:54 – The Myth of Poor AI Judgment29:20 – Final Thoughts 

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture #AISovereignty #AIInfrastructure #AIDataIn our work with clients, we already see agents doing meaningful work inside companies. Give one access to the right systems and it can search company records or prepare an order. It may send email and move a case forward without waiting for a person at every step. Far fewer companies have worked out how to let their agents deal with agents from another firm.☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture #AISovereignty #AIInfrastructure #AIDataAt first glance, the assignment looked harmless: solve a set of cybersecurity evaluation problems inside an environment with restricted internet access.[1][2] The agents hit tasks they could not complete and started looking for shortcuts. Nobody had to give them a malicious goal. They did real damage while chasing a score, because they did not inherit our norms about which methods were obviously off limits.But this hack was different. A lot different! Most serious cyberattacks are planned or coordinated by people, often teams of people. Here, the agents built the plan as they went. They found a way to leave notes, shared successful exploits across runs, and eventually reached another company☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips Chapters:00:00 – The OpenAI–Hugging Face Hack02:24 – What Actually Happened?05:40 – The Agent Alignment Problem09:15 – The Danger of Overly Broad Tool Access11:57 – Zero Trust for AI Agents14:18 – Controlling Agents with Tool Grants17:19 – Why Logging Isn’t Enough20:31 – Fighting Agents with Agents22:56 – The Future of Enterprise Cybersecurity25:20 – Building a Secure Agent Architecture27:36 – Closing Thoughts

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture #AISovereignty #AIInfrastructure #AIDataIt is still early days, but agents already participate in business processes: they accept assignments, choose tools, trigger actions, and produce results for which a person remains accountable.☑️Subscribe to our Channel for the most up-to-date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture #AISovereignty #AIInfrastructure #AIDataThe sovereignty debate has evolved in stages. Infrastructure sovereignty asks where systems run and who controls the computing substrate. Data sovereignty asks who controls the information those systems use. Intelligence sovereignty asks who controls the models that interpret that information and shape judgment. Agency sovereignty asks who controls the authority to turn judgment into action.☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

Secure Agent Series (4 of 4)#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture An agent can hold a perfectly correct authorization and still call the wrong tool, read the wrong data, or send output to the wrong place, unless something outside the agent checks every request before it executes. That check has three possible answers: yes, no, or hold. Unfortunately, most agent platforms today only have one: yes.☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture Agents now do useful work. Coding agents inspect repositories, write patches, run tests, and open pull requests. Business agents summarize accounts, triage cases, draft customer messages, prepare invoices, and move work across systems.That usefulness creates the control and governance problem. Once agents can touch repositories, ticket queues, CRM systems, email, cloud consoles, databases, and workflow tools, the enterprise needs to answer a practical question: why was this agent allowed to do this action for this task?☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips 

#AgenticAI #AIAgents #CyberSecurity #EnterpriseAI #AIArchitecture Enterprise agents are starting to do real work before many enterprises can answer a basic question: which agent did it? Some agents were never registered. Some act through user credentials or shared service accounts (for example, coding agents like Claude Code or personal agents like OpenClaw). Others have no clear owner. When an invoice is approved, a case is closed, a workflow is triggered, or an output is released, the enterprise may see the action (and potentially, the negative outcome), the tool call, or the token, but not the actor.☑️Subscribe to our Channel for the most up to date information on Agentic Mesh!You can also listen to our Podcast on Spotify: https://open.spotify.com/show/6C6U2fmVdxNMdo1bZpVASy?si=3cyoAireSZiBaKJ-1FtJlAApple Podcasts: https://podcasts.apple.com/us/podcast/the-agentic-mesh-podcast/id1874331081Out Now!  O'Reilly Agentic Mesh written by Eric Broda and Davis Broda👉https://www.oreilly.com/library/view/agentic-mesh/9798341621633/👉Stay Connected with Eric and Davis:https://www.linkedin.com/in/ericbroda/https://www.linkedin.com/in/davisbroda/👉For full-length articles by Eric Broda and Davis Broda:https://agenticmesh.substack.com/?utm_campaign=profile_chips