<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>mick delaney</title><description>Mick Delaney is a CTO, software architect and engineering leader writing and advising on production AI systems, agentic software, engineering transformation and AI governance.</description><link>https://mickdelaney.com/</link><item><title>The Protocol Firm</title><link>https://mickdelaney.com/posts/protocol-firm-1-the-protocol-firm/</link><guid isPermaLink="true">https://mickdelaney.com/posts/protocol-firm-1-the-protocol-firm/</guid><description>AI makes more intelligence available to the firm, allowing execution capacity to grow without a proportional increase in human effort at the point of execution. The Protocol Firm asks how delegated authority remains governable, testable and open to correction.</description><pubDate>Sat, 19 Sep 2026 16:00:00 GMT</pubDate></item><item><title>How celld Coordinates a Fleet Without a Separate Control Plane</title><link>https://mickdelaney.com/posts/celld-4-no-control-plane/</link><guid isPermaLink="true">https://mickdelaney.com/posts/celld-4-no-control-plane/</guid><description>How celld uses bucket leases for node discovery, failure handling and placement, and what that means for fleet credentials and network security.</description><pubDate>Wed, 16 Sep 2026 10:00:00 GMT</pubDate></item><item><title>How celld Makes SQLite Writes Durable</title><link>https://mickdelaney.com/posts/celld-3-ltx-durability/</link><guid isPermaLink="true">https://mickdelaney.com/posts/celld-3-ltx-durability/</guid><description>How celld uses LTX, peer replication and object storage to make SQLite writes durable, and which failures its recovery point objective of zero covers.</description><pubDate>Tue, 15 Sep 2026 10:00:00 GMT</pubDate></item><item><title>How celld Turns Object Storage Into a Coordination Service</title><link>https://mickdelaney.com/posts/celld-2-object-storage-coordination/</link><guid isPermaLink="true">https://mickdelaney.com/posts/celld-2-object-storage-coordination/</guid><description>celld requires only four things from an object store — conditional create, conditional overwrite, read-after-write consistency, ranged reads — and builds cell ownership, epoch fencing, and self-fencing leases on top. Here&apos;s exactly how, and what it costs.</description><pubDate>Mon, 14 Sep 2026 10:00:00 GMT</pubDate></item><item><title>celld: Durable Objects, Unbundled</title><link>https://mickdelaney.com/posts/celld-1-stateful-compute/</link><guid isPermaLink="true">https://mickdelaney.com/posts/celld-1-stateful-compute/</guid><description>celld, a new open-source project from Deno, reimplements Cloudflare&apos;s Durable Objects model on infrastructure you own — and it&apos;s a good lens for asking what should actually own an AI agent&apos;s state.</description><pubDate>Sun, 13 Sep 2026 10:00:00 GMT</pubDate></item><item><title>Why Agents Make Good Actors</title><link>https://mickdelaney.com/posts/why-agents-make-good-actors/</link><guid isPermaLink="true">https://mickdelaney.com/posts/why-agents-make-good-actors/</guid><description>AI agents are stateful, long-running, and mostly idle. That&apos;s not a quirk — it&apos;s a specification. The actor model was built for exactly this shape of work.</description><pubDate>Sun, 28 Jun 2026 10:00:00 GMT</pubDate></item><item><title>Designing environments for RL</title><link>https://mickdelaney.com/posts/rl-2-designing-environments/</link><guid isPermaLink="true">https://mickdelaney.com/posts/rl-2-designing-environments/</guid><description>Most RL writing is about algorithms. In practice, environment design — the state, action, and reward decisions that sit between your system and the algorithm — is where projects succeed or fail. This is the part that almost never gets covered.</description><pubDate>Fri, 29 May 2026 10:00:00 GMT</pubDate></item><item><title>An introduction to reinforcement learning</title><link>https://mickdelaney.com/posts/rl-1-introduction/</link><guid isPermaLink="true">https://mickdelaney.com/posts/rl-1-introduction/</guid><description>The orientation for working engineers before the rest of the series goes into algorithm internals. What RL is, the vocabulary, the Bellman idea, when to reach for it, and a map of the five posts.</description><pubDate>Fri, 29 May 2026 09:00:00 GMT</pubDate></item><item><title>Inside the Bellman loop: how value actually propagates</title><link>https://mickdelaney.com/posts/rl-3-inside-the-bellman-loop/</link><guid isPermaLink="true">https://mickdelaney.com/posts/rl-3-inside-the-bellman-loop/</guid><description>Bellman equations are usually introduced recursively, and the recursion misleads programmers. The runtime is closer to iterative graph relaxation — value sweeping backward through a state graph until it stops moving.</description><pubDate>Wed, 27 May 2026 18:00:00 GMT</pubDate></item><item><title>Inside Microsoft Agent Framework Harness Providers</title><link>https://mickdelaney.com/posts/maf-harness-providers/</link><guid isPermaLink="true">https://mickdelaney.com/posts/maf-harness-providers/</guid><description>An LLM in a loop is not enough for production work. This post breaks down the six problems every agent harness must solve — working memory, shared file access, plan/execute modes, task tracking, parallel sub-agents, and tool approval — and how the Microsoft Agents Framework implements them.</description><pubDate>Sat, 02 May 2026 10:00:00 GMT</pubDate></item><item><title>Securing MCP Pipelines with Agent Governance</title><link>https://mickdelaney.com/posts/agent-governance-mcp-pipelines/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agent-governance-mcp-pipelines/</guid><description>How to secure Model Context Protocol (MCP) tool usage in sequential agent workflows using Supply Chain Validation and Least-Privilege Sandboxing.</description><pubDate>Fri, 01 May 2026 15:00:00 GMT</pubDate></item><item><title>Securing Agent-to-Agent (A2A) Handoffs with the Agent Mesh</title><link>https://mickdelaney.com/posts/agent-governance-a2a-handoffs/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agent-governance-a2a-handoffs/</guid><description>How to secure distributed multi-agent architectures using Zero-Trust Identity, IATP encryption, and Dynamic Trust Scoring in the Agent Governance Toolkit.</description><pubDate>Fri, 01 May 2026 14:00:00 GMT</pubDate></item><item><title>Hardening AI Workflows with the Agent Governance Framework</title><link>https://mickdelaney.com/posts/agent-governance-maf-recruitment/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agent-governance-maf-recruitment/</guid><description>How to use the Microsoft Agent Framework (MAF) and Agent Governance Toolkit to build secure, governed, and Human-in-the-Loop AI workflows.</description><pubDate>Fri, 01 May 2026 10:00:00 GMT</pubDate></item><item><title>Inside the ReAct loop: how agents actually iterate</title><link>https://mickdelaney.com/posts/inside-the-react-loop/</link><guid isPermaLink="true">https://mickdelaney.com/posts/inside-the-react-loop/</guid><description>What actually happens inside one iteration of the ReAct loop — why the model is stateless, how the transcript grows, when the loop exits, and why the harness around the model is where the engineering value lives.</description><pubDate>Sun, 19 Apr 2026 10:00:00 GMT</pubDate></item><item><title>The Protocol Layer: Watching the Theory Play Out in Real Time</title><link>https://mickdelaney.com/posts/the-protocol-layer-watching-theory-play-out/</link><guid isPermaLink="true">https://mickdelaney.com/posts/the-protocol-layer-watching-theory-play-out/</guid><description>Google&apos;s developer guide to AI agent protocols is framed as a tutorial. Read it through a Coasean lens and it&apos;s something more interesting — live evidence that the transaction costs making firms necessary are being eliminated, layer by layer, in the open.</description><pubDate>Fri, 03 Apr 2026 10:00:00 GMT</pubDate></item><item><title>The Firm in 2026</title><link>https://mickdelaney.com/posts/firm-under-ai-16-the-firm-in-2026/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-16-the-firm-in-2026/</guid><description>Returning to Coase&apos;s original question — why do firms exist? — with fifteen posts of new machinery. The answer has not changed. The composition of the costs has.</description><pubDate>Tue, 17 Mar 2026 10:00:00 GMT</pubDate></item><item><title>The Agentic Ecosystem</title><link>https://mickdelaney.com/posts/firm-under-ai-15-the-agentic-ecosystem/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-15-the-agentic-ecosystem/</guid><description>The shift from hierarchy to protocol-governed agent coordination is not theoretical. Over fifty companies are engineering it now. Here is what the early evidence tells us.</description><pubDate>Sun, 15 Mar 2026 10:00:00 GMT</pubDate></item><item><title>Value Stream Mapping</title><link>https://mickdelaney.com/posts/firm-under-ai-14-value-stream-mapping/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-14-value-stream-mapping/</guid><description>Protocols define how participants interact. Value streams define why. By mapping the flow of outcomes rather than the structure of teams, we align architecture, organisation, and network around the same thing: how value is actually produced.</description><pubDate>Fri, 13 Mar 2026 07:00:00 GMT</pubDate></item><item><title>The Physics of Flow</title><link>https://mickdelaney.com/posts/firm-under-ai-13-the-physics-of-flow/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-13-the-physics-of-flow/</guid><description>Queuing theory reveals the hidden physics governing how work flows through organisations. Every boundary is a queue. High utilisation destroys responsiveness. And AI agents change the service rate — but not the dynamics.</description><pubDate>Thu, 12 Mar 2026 10:00:00 GMT</pubDate></item><item><title>The Protocol Layer</title><link>https://mickdelaney.com/posts/firm-under-ai-12-the-protocol-layer/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-12-the-protocol-layer/</guid><description>How networked firms coordinate work through protocols instead of hierarchy.</description><pubDate>Wed, 11 Mar 2026 10:00:00 GMT</pubDate></item><item><title>The Geometry of Networks</title><link>https://mickdelaney.com/posts/firm-under-ai-11-the-geometry-of-networks/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-11-the-geometry-of-networks/</guid><description>Not all networks scale the same way. The geometry of a network — broadcast, connection, or group-forming — determines how its value grows, and which institutions it produces.</description><pubDate>Mon, 09 Mar 2026 12:00:00 GMT</pubDate></item><item><title>Metcalfe&apos;s Law, Conway&apos;s Law, and the Networked Firm</title><link>https://mickdelaney.com/posts/firm-under-ai-10-metcalfes-law-conways-law-networked-firm/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-10-metcalfes-law-conways-law-networked-firm/</guid><description>Metcalfe&apos;s Law governs the value of the external network, while Conway&apos;s Law governs the structure of the organization that builds and operates it.</description><pubDate>Sun, 08 Mar 2026 10:00:00 GMT</pubDate></item><item><title>The Hybrid Topology</title><link>https://mickdelaney.com/posts/firm-under-ai-9-the-hybrid-topology/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-9-the-hybrid-topology/</guid><description>We are witnessing the collapse of the corporate Tree into the Hybrid Mesh. As intelligence becomes infrastructure, the distance between strategy and execution shrinks to a single, high-density hop.</description><pubDate>Sat, 07 Mar 2026 10:00:00 GMT</pubDate></item><item><title>Knowledge Diffusion and Network Density</title><link>https://mickdelaney.com/posts/firm-under-ai-8-knowledge-diffusion-and-network-density/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-8-knowledge-diffusion-and-network-density/</guid><description>Firms do not only coordinate production. They concentrate knowledge. Innovation has historically depended on short-distance diffusion through dense networks.</description><pubDate>Sun, 22 Feb 2026 10:00:00 GMT</pubDate></item><item><title>Agentic Labour: When Work Becomes Executable</title><link>https://mickdelaney.com/posts/firm-under-ai-7-agentic-labour-when-work-becomes-executable/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-7-agentic-labour-when-work-becomes-executable/</guid><description>Software first reduced coordination costs. Now autonomous agents begin to execute parts of labor itself. This changes the structure of work without eliminating it.</description><pubDate>Fri, 20 Feb 2026 18:30:00 GMT</pubDate></item><item><title>The Platform as Proto-Firm</title><link>https://mickdelaney.com/posts/firm-under-ai-6-the-platform-as-proto-firm/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-6-the-platform-as-proto-firm/</guid><description>Platforms are not the disappearance of the corporation. They are a reconfiguration of how coordination occurs within a corporate shell.</description><pubDate>Thu, 19 Feb 2026 20:15:00 GMT</pubDate></item><item><title>The Boundary of the Firm in a Digital Age</title><link>https://mickdelaney.com/posts/firm-under-ai-5-the-boundary-of-the-firm-in-a-digital-age/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-5-the-boundary-of-the-firm-in-a-digital-age/</guid><description>If firms exist because coordination is costly, what happens when coordination becomes cheaper again? The boundary of the firm is not fixed.</description><pubDate>Wed, 18 Feb 2026 18:30:00 GMT</pubDate></item><item><title>The Innovator&apos;s Dilemma</title><link>https://mickdelaney.com/posts/firm-under-ai-4-the-innovators-dilemma/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-4-the-innovators-dilemma/</guid><description>When coordination costs shift, firm boundaries should adjust. But mature corporations are optimised for a cost structure that no longer holds — and their own incentives prevent adaptation.</description><pubDate>Tue, 17 Feb 2026 20:15:00 GMT</pubDate></item><item><title>When Corporations Became Too Big</title><link>https://mickdelaney.com/posts/firm-under-ai-3-when-corporations-became-too-big/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-3-when-corporations-became-too-big/</guid><description>The managerial corporation solved coordination at scale. But hierarchy has structural limits — and the same mechanisms that enabled growth eventually constrained it.</description><pubDate>Mon, 16 Feb 2026 19:15:00 GMT</pubDate></item><item><title>The Rise of the Managerial Corporation</title><link>https://mickdelaney.com/posts/firm-under-ai-2-rise-of-the-managerial-corporation/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-2-rise-of-the-managerial-corporation/</guid><description>Corporations did not just reduce transaction costs. In the 20th century, they became the dominant institution for organising production, concentrating capital, and coordinating labour at scale.</description><pubDate>Sun, 15 Feb 2026 18:30:00 GMT</pubDate></item><item><title>Why Corporations Exist: A Coasean Foundation</title><link>https://mickdelaney.com/posts/firm-under-ai-1-why-corporations-exist-a-coasean-foundation/</link><guid isPermaLink="true">https://mickdelaney.com/posts/firm-under-ai-1-why-corporations-exist-a-coasean-foundation/</guid><description>Corporations did not emerge by accident. They are coordination machines. In 2026, as transaction costs fall, we revisit why they exist — and what might be changing.</description><pubDate>Sat, 14 Feb 2026 10:00:00 GMT</pubDate></item><item><title>Metrics Are Projection</title><link>https://mickdelaney.com/posts/metrics-are-projection/</link><guid isPermaLink="true">https://mickdelaney.com/posts/metrics-are-projection/</guid><description>Metrics don&apos;t measure reality — they project it. As systems become more autonomous, projection design matters more than model intelligence.</description><pubDate>Mon, 09 Feb 2026 10:00:00 GMT</pubDate></item><item><title>Agent Experimentation and Innovation at the Edge</title><link>https://mickdelaney.com/posts/agent-experimentation-at-the-edge/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agent-experimentation-at-the-edge/</guid><description>Execution, collaboration, and ecosystems: three edges where agent innovation is actually happening — and what it means for building durable, human-centered systems.</description><pubDate>Wed, 04 Feb 2026 10:00:00 GMT</pubDate></item><item><title>From Process Managers to Stable Agent Workflows</title><link>https://mickdelaney.com/posts/agent-stability/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agent-stability/</guid><description>Agent workflows feel new, but the stability problems they face are not. Here&apos;s how decades of integration architecture apply to building production agent systems.</description><pubDate>Sat, 24 Jan 2026 10:00:00 GMT</pubDate></item><item><title>Keeping State Consistent: Database Transactions in LangGraph Workflows</title><link>https://mickdelaney.com/posts/langgraph-database-transactions/</link><guid isPermaLink="true">https://mickdelaney.com/posts/langgraph-database-transactions/</guid><description>Three patterns for keeping LangGraph workflow state and database state synchronized - idempotent tasks, read-before-write synchronization, and the transactional outbox pattern.</description><pubDate>Thu, 22 Jan 2026 18:00:00 GMT</pubDate></item><item><title>From Data to Decisions: The Statistical Foundations of Experimentation</title><link>https://mickdelaney.com/posts/data-to-decisions/</link><guid isPermaLink="true">https://mickdelaney.com/posts/data-to-decisions/</guid><description>Building intuition for the statistics behind every A/B test — from raw observations to principled decisions, one concept at a time.</description><pubDate>Mon, 17 Nov 2025 10:00:00 GMT</pubDate></item><item><title>Maths-as-Code for Machine Learning</title><link>https://mickdelaney.com/posts/maths-as-code-for-ml/</link><guid isPermaLink="true">https://mickdelaney.com/posts/maths-as-code-for-ml/</guid><description>Translate the maths you see in ML papers into code you can write. A compact cheat sheet mapping notation → intuition → Python/NumPy (and JS).</description><pubDate>Mon, 03 Nov 2025 00:00:00 GMT</pubDate></item><item><title>Agents, Routing, Patterns, and Actors</title><link>https://mickdelaney.com/posts/agents-routing-patterns-and-actors/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agents-routing-patterns-and-actors/</guid><description>Building Agent Networks that Don&apos;t Collapse Under Their Own Cleverness</description><pubDate>Tue, 17 Jun 2025 10:00:00 GMT</pubDate></item><item><title>Agents Are Still Just Software</title><link>https://mickdelaney.com/posts/agents-are-still-just-software/</link><guid isPermaLink="true">https://mickdelaney.com/posts/agents-are-still-just-software/</guid><description>Agents Are Just Software Why the Next Wave of AI Is Really a Distributed-Systems Story.</description><pubDate>Sun, 08 Jun 2025 10:00:00 GMT</pubDate></item></channel></rss>