• Three Companies Now Run Your AI. Could You Leave Any of Them?

    2026-08-15 | 12 minute read

    Executive Summary

    Three foundation model providers now control roughly 90% of enterprise AI spend, and only 6% of leaders believe they could switch their main one without disruption. In Europe, that dependency has stopped being a procurement preference and become a supervised risk. A consultant's guide for leaders who bought capability and acquired a dependency they never priced.

  • The Seniority Cliff: You Automated the Work That Made Your Experts

    2026-07-31 | 11 minute read

    Executive Summary

    Junior employment is falling at AI-adopting firms while senior employment holds. The tasks being automated were never valuable in themselves. They were how judgment got transferred. A consultant's guide for leaders who are optimising a five-year cost line against a ten-year capability line.

  • You Are Deploying AI. Your Competitor Is Shipping It.

    2026-07-15 | 12 minute read

    Executive Summary

    Almost every enterprise AI programme in Europe is pointed inward: copilots, back-office workflows, the cost line. For companies that sell physical products, the profit pool sits in the service business, and three European regulations are rewriting the rules around it while the AI programme summarises meetings. A consultant's guide for leaders whose AI strategy is really an IT strategy.

  • Your AI Strategy Is a Data Strategy in Disguise

    2026-06-30 | 9 minute read

    Executive Summary

    Only 7% of enterprises say their data is ready for AI. The other 93% are running transformation programmes on a foundation they never built. A consultant's guide to the layer nobody puts on a board slide, and why in Europe it has quietly become a legal precondition rather than an engineering nicety.

  • The Verification Bottleneck: When Human Oversight Becomes the Thing That Cannot Scale

    2026-06-15 | 8 minute read

    Executive Summary

    Enterprises spent two years making AI agents dramatically more capable. Almost none scaled the human capacity to check what those agents produce. As autonomy rises, the binding constraint quietly moves from execution to verification, and in Europe that constraint is about to become a legal obligation. A consultant's guide for leaders who do not want oversight to become the bottleneck they never budgeted for.

  • AI Is Eating Your Org Chart: What Leaders Get Wrong About Restructuring Around AI

    2026-05-31 | 11 minute read

    Executive Summary

    PayPal is cutting a fifth of its workforce. Cloudflare eliminated 1,100 roles and called it an agentic AI-first operating model. The headlines say AI is replacing people. The reality is more uncomfortable: most of these companies are cutting the org chart before they have redesigned the work, and before they have built the governance to run what is left. A consultant's guide for leaders who do not want to be the cautionary tale.

  • The AI-Native Competitor Is Already In Your Market

    2026-05-15 | 10 minute read

    Executive Summary

    In 2024, AI-native startups captured 36% of the enterprise AI application market. In 2025, they captured 63%. The line was crossed quietly, and most incumbent boards still don't know it has happened. A consultant's guide for leaders trying to work out whether they are the disruptor or the disrupted — and what to do before the next budget cycle.

  • “We Bought the Copilot. Why Is Nothing Happening?”

    2026-04-30 | 6 minute read

    Executive Summary

    The meeting usually happens sometime in the fourth quarter. The CFO opens the invoice, multiplies thirty dollars a month by the number of seats, and asks the CIO the question that should have been asked eighteen months ago: “What are we actually getting from this?” In a 5,000-seat enterprise, a full Microsoft 365 Copilot deployment runs roughly $1.8 million a year before training, change management, or integration costs. This is the activation problem — and in 2026, it is the single largest source of invisible AI waste in the enterprise.

  • “We Trained 5,000 People on AI. Nothing Changed.”

    2026-04-15 | 7 minute read

    Executive Summary

    This article explains why large-scale enterprise AI training efforts often fail to translate into operational change. It identifies four structural failure modes: generic content, training detached from daily tools, insufficient manager enablement, and outcome measurement focused on completions instead of behaviour. It proposes a practical playbook: role-specific, in-context training tied to workflow redesign, with clear before-and-after behaviour and business metrics. The central argument is that training without redesign is cost, not capability.

  • Your Shadow AI Policy Is Already Being Written — Just Not By You

    2026-03-31 | 9 minute read

    Executive Summary

    This article explains why shadow AI is now an operating reality, not an edge case, and why bans fail to reduce risk. It outlines a practical governance approach: create visibility, provide a safer and better approved default, and define clear function-level boundaries. The key message for senior leaders is that unmanaged AI behaviour is already writing policy by default, and only decisive leadership can convert hidden exposure into controlled productivity.

  • Time Saved Is Not Value Captured: What Leaders Miss About AI Productivity

    2026-03-15 | 9 minute read

    Executive Summary

    This article explains why time saved by AI rarely becomes enterprise value without workflow redesign, explicit usage boundaries, and CFO-grade outcome measurement. It outlines three leadership moves and a practical 90-day playbook to translate task-level productivity into measurable P&L impact.

  • From Pilots to Production: What the Companies That Actually Scaled AI Did Differently

    2026-02-28 | 9 minute read

    Executive Summary

    This series has covered strategy, economics, and architecture. What it hasn't provided yet is evidence: what does production-scale AI success actually look like? This post closes that gap. Three recent enterprise deployments - Walmart, Deutsche Telekom, and a global pharmaceutical firm - made it from pilot to production at scale. The patterns they share are not what most AI strategies are built around. They targeted unglamorous workflows instead of strategic priorities. They designed human oversight in from the start rather than adding it as a safeguard afterward. And they reduced operational burden without transferring accountability. These three principles form a production readiness test - one you can apply before signing off on your next scale investment.

  • Scaling Agentic AI: The 2026 Enterprise Blueprint

    2026-02-15 | 7 minute read

    Executive Summary

    This article presents a practical enterprise blueprint for scaling agentic AI in 2026. It argues that the market has moved beyond experimentation: while adoption intent is high, production-scale success remains limited because most programs underestimate operational complexity. The post defines agentic AI as systems that can plan, act through tools, adapt to outcomes, and retain context, then explains why initiatives break down across five recurring pressure points: reliability risk, cost escalation, latency constraints, governance gaps, and integration debt. To address this, it introduces a five-layer architecture covering interfaces, specialized agents, orchestration, tooling, and governance controls, with human oversight as a core design principle. It also provides a build-versus-buy framework and highlights high-value use cases in healthcare, customer operations, and supply chain environments. The core message is that sustainable value will not come from better prompts alone, but from robust system architecture, controllability, observability, and disciplined operating models that let enterprises scale agents safely and economically.

  • The Hidden Cost of AI: Why Even the Right Use Cases Fail to Deliver ROI

    2026-01-31 | 8 minute read

    Executive Summary

    Many executive teams now choose technically feasible AI use cases with clear KPIs, yet financial outcomes still disappoint. This article explains why: most ROI leakage happens after a pilot looks successful. Hidden costs in data cleanup, legacy integration, model monitoring, compliance work, and change management often compound faster than expected savings. It shows how seemingly successful pilots still fail at scale when labor is reconfigured instead of reduced, throughput bottlenecks remain untouched, and quality risks create downstream rework. The piece also separates activity metrics from true financial outcomes by emphasizing CFO-grade measurement tied to EBITDA, margin, and cash impact. The practical takeaway is a pre-scale stress-test: map full total cost of ownership, define error budgets, verify adoption fit, and prove value in short cycles. The core argument is that sustainable AI returns come from systems redesign, not simple tool deployment.

  • Finding the AI Use Cases That Actually Matter - A Consultant's Guide for Enterprise Leaders

    2026-01-15 | 13 minute read

    Executive Summary

    This article gives enterprise leaders a practical method for identifying AI initiatives that actually create business value. Instead of starting with tools, it starts with decision bottlenecks: repeated, high-stakes choices where better speed or accuracy affects revenue, cost, risk, or cycle time. It introduces a three-filter model for prioritization: business value, execution feasibility, and time to measurable impact. The post extends this with a value-feasibility matrix and a governance checklist to reduce wasted spend before projects begin. A central focus is avoiding pilot purgatory, where teams produce promising demos that never scale due to weak ownership, poor data readiness, and missing scale criteria. It recommends explicit scale-or-kill gates, short pilot windows, and business-owner accountability to keep portfolios focused. The core message is clear: AI success is not about running more experiments, but about selecting fewer, better use cases with strong economics and a realistic path to adoption.