Insights.
Research, operating models, and field notes for industrial B2B leaders trying to turn AI from activity into accountable business change.
On this page
- Your AI Saved an Hour. Who Gets It?
AI can make everyone faster while the customer waits just as long. The missing decision is what happens to the capacity you free: which queue moves, which cost disappears, and who owns the result.
- Write the Review Before the Agent Starts
An agent can finish a task before you have decided what finished means. A better brief makes the evidence, acceptance checks and permission boundaries clear from the start.
- Is This Already AGI? The Better Question Is What You Hand Over
Replace the AGI naming debate with a local test: complete cases, permitted actions, review effort and the people who can recognise a consequential error.
- The Verification Ceiling: When AI Produces Faster Than You Can Check
AI can generate a quotation faster than the company can defend its combined promise. Match required checks, review capacity and release authority to the consequence.
- Your First AI Profit May Be in the Order Book
AI demand is reaching power, transformers, controls, automation, and semiconductors. Test the opportunity with live customer evidence before funding it.
- Your AI Sandbox Can Have Customers on the Other Side
A preliminary evaluation incident makes an access question concrete: what can a persistent system reach indirectly, and who can stop it before a test has external consequences?
- EU AI Act Article 50: The Record Behind the Label
A disclosure review and a commercial claim record answer different questions. Trace one approved product claim through evidence, review, translation, release and correction.
- The Button Test: What Your AI Agent Actually Completes
A human approval can be a deliberate control. Map the autonomous steps, retained decisions and completed outcome before funding an agent workflow.
- After Scarcity: The Work of Living Well
A philosophical exploration of AI abundance, distribution and human agency. Material security matters; it does not decide who receives the gains or what people can choose to do.
- Enterprise AI: Where Value Survives the Full Cost
Judge enterprise AI through completed outcomes, full delivered cost and evidence for scaling. Adoption and model prices are useful signals; the workflow still needs its own business case.
- AI-Assisted B2B Content: Give the Buyer a Reason to Care
Interchangeable copy fails the reader before a style detector enters the conversation. Build B2B content around a useful decision, applicable evidence and accountable judgment.
- The Generative AI Reset: Marketing Beyond the Click
Examine what buyers encounter in AI answers, then repair a specific evidence or ownership gap. A repeatable query baseline belongs beside visits, enquiries and commercial outcomes.
- AI Cold Outreach Needs a Reason to Be Sent
AI can lower the cost of sales research and drafting. Leaders still need an accountable contact decision, verifiable account relevance and a scorecard that includes recipient feedback.
- Scale AI Content Only as Fast as You Can Stand Behind It
Product guidance, localisation and executive commentary need different judgments. Design their ownership, evidence, review and economics before expanding AI production.
- AI Changes the SaaS Renewal Decision
Cheaper software creation expands build-versus-buy options. Compare subscriptions, internal services and agentic layers on the same horizon, with ownership and transition costs included.
- State of AI, April 2026: Three Decisions to Make Carefully
An archived April 2026 executive memo on provider choice, longer delegated work and capacity commitments, with evidence and reversal conditions for each.
- The Intelligent Industrial Enterprise: Control Before Scale
A planning framework for industrial AI: connect process outcomes to decision rights, supplier continuity and workforce readiness, then use conditional 2036 scenarios to test investment choices.
- Computational Scale Matters When It Changes the Work
Translate hardware, algorithm and compute progress into accepted-task economics, service reliability and investment conditions instead of a universal automation forecast.
- Find AI Opportunities Through Business Processes
Start with a process and a performance gap. Compare AI with process repair, integration and rules, then choose the next experiment using evidence, delivery conditions and realistic economics.
- When AI Pilots Become Procrastination
A useful pilot resolves an uncertainty that can change an investment decision. Design its evidence, stage and commitments around that purpose rather than a production deadline.
- Define the AI Transformation Office Before You Hire It
Create a central team for a named coordination problem, with delegated decision rights and a funded receiving organisation. Transfer work when readiness is demonstrated.
- Design the CAIO Mandate Around Its Decisions
Choose the operating problem before the executive title. Operator, builder, reformer and federator mandates need different authority, partners and measures.
- Human-in-the-Loop Needs More Than an Approve Button
Design AI oversight around the error it must detect and the consequence it must prevent. Queue volume and agreement between models cannot establish that a release is sound.
- Start the AI Programme With the Operating Problem
An opportunity inventory is not an investment portfolio. Diagnose a recurring process problem, compare interventions and fund the evidence that can change the next decision.
- An AI Can Produce a Campaign. Running an Agency Takes More.
A fast agency demo changes the cost of producing a first campaign. It leaves a harder question unanswered: who owns performance, client trust and the work that continues after launch?
- AI Capability Can Improve While AI Investments Disappoint
Benchmark progress, reliable operating performance and investment returns are different questions. An executive needs a plan that can respond to capability gains without depending on a forecast of stock prices or AGI.
- What It Takes to Hand a Workflow to an AI Agent
The important unit of AI orchestration is a handoff. A worked quotation example shows how decision rights, evidence, permissions and recovery determine whether a faster task becomes a dependable process.
- Who Owns an AI System After It Goes Live?
An AI operating model becomes useful when it assigns the decisions, maintenance and costs that continue after the launch.
- Vibe Coding Needs an Owner After the Demo
AI can help a domain expert make an idea tangible. The enterprise decision is who can inspect, maintain and recover the software once people depend on it.
- How Far Can an Industrial AI Pilot Support the Next Investment?
A promising ordering pilot can justify a next step. Scaling requires evidence about what changes in the next customer group, product range or operating boundary.
- Make B2B Evidence Useful to Buyers and Their AI Assistants
Clear, inspectable evidence helps a buyer evaluate you and makes your business easier for an AI-assisted search to describe accurately.
- AI Applications Need an Outcome You Can Trace
Support, forecasting, document review and lead prioritization can all be useful. Their business cases depend on the decisions and work after the model responds.
- The Question an AI Business Case Has to Answer
A CFO’s question changed how I select AI work: explain the value, check the conditions and fund uncertainty deliberately.
- The Workforce Question Behind an AI Rollout
People ask about training, but the harder questions concern identity, authority and what leadership can honestly promise.
- What Three Years of AI Writing Actually Saved Me
An email I nearly sent showed me where AI saves effort—and where my own judgment still has to do the work.
- Your Sales Incentives Shape the Customer Handoff
Customer satisfaction becomes a commercial responsibility when compensation, commitments and the handoff extend beyond the signed contract.
- How I Use NotebookLM Before a Board Discussion
Listening to a research summary helps me get oriented. The board paper still needs claims traced to their sources and judgment I can explain.