Lucas Northhttps://lucasnorth.ukClear thinking about complex technology.The Architecture of a Trustworthy AI Workflowhttps://lucasnorth.uk/articles/architecture-trustworthy-ai-workflowTrust emerges from evidence, bounded actions, explicit decisions and recoverable execution across the whole workflow.Wed, 08 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/architecture-trustworthy-ai-workflowThe Case Against Building an Enterprise Chatbothttps://lucasnorth.uk/articles/case-against-enterprise-chatbotA universal chat interface often shifts the work of understanding systems onto users. Focused workflow products usually create more value.Wed, 08 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/case-against-enterprise-chatbotYour Data Is Not Ready for AI, and That’s Finehttps://lucasnorth.uk/articles/data-not-ready-for-aiData readiness is not a binary gate. Start with the sources and quality level required by one valuable decision.Wed, 08 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/data-not-ready-for-aiThe Power Platform Centre of Excellence Starter Kit Is Not a Governance Strategyhttps://lucasnorth.uk/articles/power-platform-coe-kit-not-governanceThe CoE Starter Kit provides useful telemetry and workflows, but it cannot decide what your organisation should govern or why.Wed, 08 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-platform-coe-kit-not-governanceYour Power Platform Estate Is Already a Software Portfoliohttps://lucasnorth.uk/articles/power-platform-estate-software-portfolioPower Platform workloads deserve portfolio decisions about investment, ownership, duplication, risk and retirement.Wed, 08 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-platform-estate-software-portfolioAI Agents Should Earn Their Permissionshttps://lucasnorth.uk/articles/agents-earn-permissionsAgent autonomy should expand through evidence, with permissions tied to demonstrated reliability and consequence.Tue, 07 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/agents-earn-permissionsWhat Good Power Platform Governance Looks Like in Practicehttps://lucasnorth.uk/articles/good-power-platform-governance-practiceGood governance combines clear boundaries, proportionate controls, usable delivery paths and accountable service ownership.Tue, 07 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/good-power-platform-governance-practiceHuman-in-the-Loop Is Not a Complete Safety Strategyhttps://lucasnorth.uk/articles/human-in-loop-incompleteHuman review only reduces risk when reviewers have authority, evidence, time and a realistic chance of detecting failure.Tue, 07 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/human-in-loop-incompleteThe Power Platform Environment Strategy Most Organisations Actually Needhttps://lucasnorth.uk/articles/practical-power-platform-environment-strategyA practical environment model separates personal productivity, team experimentation and lifecycle-managed production.Tue, 07 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/practical-power-platform-environment-strategyStop Asking Whether the Model Is Accuratehttps://lucasnorth.uk/articles/stop-asking-model-accurateAccuracy without a defined task, distribution and consequence is not a useful production measure.Tue, 07 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/stop-asking-model-accurateAI Agents Need Boundaries, Not Better Promptshttps://lucasnorth.uk/articles/ai-agents-need-boundariesReliable agents come from permissions, validation and controlled execution, not increasingly elaborate instructions.Thu, 09 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-agents-need-boundariesWhy Every AI System Needs a Refusal Strategyhttps://lucasnorth.uk/articles/ai-refusal-strategyA useful refusal is a designed product outcome that protects users and moves uncertain work somewhere responsible.Thu, 09 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-refusal-strategyWhy Your Copilot Rollout Has Become a Content Audithttps://lucasnorth.uk/articles/copilot-rollout-content-auditCopilot exposes years of unresolved ownership, permissions and lifecycle decisions because its answers can only be as trustworthy as its sources.Thu, 09 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/copilot-rollout-content-auditEnterprise AI Fails at the Handoffshttps://lucasnorth.uk/articles/enterprise-ai-fails-at-handoffsThe model may perform its task well while the wider system fails between people, tools, decisions and exception queues.Thu, 09 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/enterprise-ai-fails-at-handoffsRetrieval-Augmented Generation Is a Data Producthttps://lucasnorth.uk/articles/rag-is-data-productRAG succeeds through owned sources, measurable retrieval and disciplined lifecycle management.Thu, 09 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/rag-is-data-productAI Observability Is More Than Token Countshttps://lucasnorth.uk/articles/ai-observability-more-than-tokensUseful observability connects requests, evidence, decisions, actions and outcomes without retaining data indiscriminately.Sun, 12 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-observability-more-than-tokensMicrosoft Copilot Isn't Magic. It's Architecturehttps://lucasnorth.uk/articles/copilot-architectureCopilot experiences are shaped by identity, permissions, retrieval and integration architecture long before a prompt reaches the model.Sun, 12 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/copilot-architectureWhy Prompt Engineering Won’t Save a Broken Workflowhttps://lucasnorth.uk/articles/prompt-engineering-broken-workflowPrompts can shape model behaviour, but they cannot repair missing evidence, unclear ownership or wasteful handoffs.Sun, 12 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/prompt-engineering-broken-workflowWhy Most Enterprise AI Projects Fail (And How to Avoid It)https://lucasnorth.uk/articles/why-enterprise-ai-projects-failEnterprise AI projects fail when compelling demos are mistaken for operational systems. The remedy is a narrow outcome, owned data and production evidence.Sun, 12 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/why-enterprise-ai-projects-failA Cloud Flow Is Not a Business Processhttps://lucasnorth.uk/articles/cloud-flow-not-business-processA flow implements part of a process; it does not contain the ownership, exceptions and outcomes that make the process operable.Sun, 05 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/cloud-flow-not-business-processDelegation Is an Architecture Constraint, Not a Warning to Ignorehttps://lucasnorth.uk/articles/delegation-architecture-constraintNon-delegable formulas can silently produce incomplete results, making data-source choice and query design correctness issues.Sun, 05 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/delegation-architecture-constraintWhy Long-Running Power Automate Flows Become Operational Riskshttps://lucasnorth.uk/articles/long-running-flows-operational-riskExtended flows accumulate state, dependency and recovery problems that run history alone cannot manage.Sun, 05 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/long-running-flows-operational-riskPower Automate Needs Idempotencyhttps://lucasnorth.uk/articles/power-automate-idempotencyRetries and duplicate triggers are normal; flows must prevent the same business action from happening twice.Sun, 05 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-automate-idempotencyA Successful Flow Run Does Not Mean the Business Process Succeededhttps://lucasnorth.uk/articles/successful-flow-run-not-business-successTechnical completion and business outcome require separate states and measures.Sun, 05 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/successful-flow-run-not-business-successAI Governance Doesn't Have to Kill Innovationhttps://lucasnorth.uk/articles/ai-governance-innovationThe best AI governance creates clear, fast paths for experimentation while reserving heavier controls for decisions that can cause real harm.Sat, 11 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-governance-innovationChoosing Between SharePoint, Dataverse and SQL for Power Appshttps://lucasnorth.uk/articles/choose-sharepoint-dataverse-sqlChoose storage from required behaviour, security and ownership, not familiarity alone.Sat, 11 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/choose-sharepoint-dataverse-sqlWhat Production-Ready AI Actually Looks Likehttps://lucasnorth.uk/articles/production-ready-aiProduction readiness is not a model benchmark. It is the ability to operate, explain, recover and improve a complete AI system.Sat, 11 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/production-ready-aiStop Measuring AI Adoption by Licence Activationhttps://lucasnorth.uk/articles/stop-measuring-ai-adoptionActivation and prompt counts measure access and curiosity. Leaders need evidence that AI changes valuable work.Sat, 11 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/stop-measuring-ai-adoptionAI Governance Should Be an Engineering Systemhttps://lucasnorth.uk/articles/ai-governance-engineering-systemGovernance becomes effective when policies compile into controls that can be tested, observed and evidenced in running systems.Sat, 04 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-governance-engineering-systemThe useful constraint in edge-first SaaS architecturehttps://lucasnorth.uk/articles/cloudflare-saas-boundariesCloudflare’s runtime rewards explicit boundaries. That constraint can produce simpler systems when state and background work are designed deliberately.Sat, 04 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/cloudflare-saas-boundariesThe Dangerous Gap Between Connector Access and Data Authorisationhttps://lucasnorth.uk/articles/connector-access-data-authorisation-gapA permitted connector does not mean the user or flow should access every operation and record behind it.Sat, 04 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/connector-access-data-authorisation-gapThe Default Environment Is Not a Development Environmenthttps://lucasnorth.uk/articles/default-environment-not-developmentThe default environment is a shared productivity space, not a safe foundation for lifecycle-managed business applications.Sat, 04 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/default-environment-not-developmentData Loss Prevention Is Not the Same as Data Governancehttps://lucasnorth.uk/articles/dlp-not-data-governanceDLP limits connector combinations; data governance defines meaning, ownership, access, quality and lifecycle.Sat, 04 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/dlp-not-data-governancePower Pages Is a Security Boundary, Not Just a Website Builderhttps://lucasnorth.uk/articles/power-pages-security-boundaryExternal identity, table permissions and web roles require deliberate design because the audience sits outside the tenant.Sat, 04 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-pages-security-boundaryWhy most Copilot pilots never become operating systemshttps://lucasnorth.uk/articles/copilot-pilot-to-productionThe technical demo is rarely the hard part. Production adoption depends on permissions, content ownership and an operating model that survives the launch meeting.Mon, 13 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/copilot-pilot-to-productionCopilot Can’t Fix Your Information Architecturehttps://lucasnorth.uk/articles/copilot-cannot-fix-information-architectureCopilot can improve access to knowledge, but it cannot decide which source is authoritative or who should own it.Mon, 06 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/copilot-cannot-fix-information-architectureThe Graduation Path Every Power Platform App Needshttps://lucasnorth.uk/articles/power-platform-graduation-pathWorkloads need a predictable route from personal experiment to supported business service.Mon, 06 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-platform-graduation-pathOwnership Is the Hardest Power Platform Governance Problemhttps://lucasnorth.uk/articles/power-platform-ownership-hardest-problemTechnical ownership, business accountability and connection identity are different responsibilities that inventories often collapse.Mon, 06 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-platform-ownership-hardest-problemA Risk-Based Governance Model for Power Platformhttps://lucasnorth.uk/articles/risk-based-power-platform-governanceRisk-based governance links observable workload characteristics to proportionate platform and operating controls.Mon, 06 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/risk-based-power-platform-governancePower Platform Solutions Should Be the Unit of Ownershiphttps://lucasnorth.uk/articles/solutions-unit-of-ownershipSolutions align ownership with deployment, dependencies and the business change being delivered.Mon, 06 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/solutions-unit-of-ownershipAI governance needs a control plane, not another committeehttps://lucasnorth.uk/articles/ai-governance-control-planePolicies become useful only when they are translated into enforceable decisions across identity, data, models and runtime behaviour.Fri, 10 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-governance-control-planeYour AI Strategy Is Probably Just a Vendor Listhttps://lucasnorth.uk/articles/ai-strategy-vendor-listA sequence of product purchases is not a strategy. Real AI strategy makes choices about advantage, operating capability and where not to invest.Fri, 10 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/ai-strategy-vendor-listThe Hidden Architecture Behind Every AI Assistanthttps://lucasnorth.uk/articles/hidden-architecture-ai-assistantThe conversational interface is the visible edge of a system spanning identity, retrieval, policy, tools, telemetry and human escalation.Fri, 10 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/hidden-architecture-ai-assistantPower Platform Business Continuity Starts Before the Maker Leaveshttps://lucasnorth.uk/articles/power-platform-continuity-before-maker-leavesContinuity comes from designed ownership, deployability and recovery, not emergency transfer during offboarding.Fri, 10 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/power-platform-continuity-before-maker-leavesThe Real Cost of Enterprise AI Is Operationalhttps://lucasnorth.uk/articles/real-cost-enterprise-aiModel tokens are visible, but integration, verification, support, data ownership and change management dominate many enterprise AI costs.Fri, 10 Jul 2026 00:00:00 GMThttps://lucasnorth.uk/articles/real-cost-enterprise-ai