As enterprises invest more in AI, executives expect measurable financial returns. But not every AI investment can be reliably quantified in monetary terms. This article explores what can realistically be measured, what remains uncertain, and why executives need to understand these limits.

As enterprises invest more in AI, executives expect measurable financial returns. But not every AI investment can be reliably quantified in monetary terms. This article explores what can realistically be measured, what remains uncertain, and why executives need to understand these limits.

MCP gives external AI controlled access to the software and systems knowledge workers already use. This article helps knowledge workers understand the available ways AI can work with software, choose the right approach for a task, and set appropriate permissions so they can gain more practical help from AI without losing oversight.

MCP gives external AI controlled access to the software and systems knowledge workers already use. This article helps knowledge workers understand the available ways AI can work with software, choose the right approach for a task, and set appropriate permissions so they can gain more practical help from AI without losing oversight.

This article aims to provide C-suite leaders a practical foundation for deciding how AI should be used or implemented within their enterprise. It explains what AI can do, where its outputs become harder to trust, and why human judgement and responsibility must remain central.

บทความนี้ประสงค์ที่จะวางรากฐานความเข้าใจเกี่ยวกับ AI ให้แก่ผู้บริหาร เพื่อใช้ประกอบการตัดสินใจว่าควรนำ AI มาใช้ในองค์กรอย่างไร โดยอธิบายว่า AI ทำอะไรได้บ้าง ผลลัพธ์ของ AI เชื่อถือได้ในสถานการณ์ใด และเหตุใดวิจารณญาณและความรับผิดชอบของมนุษย์จึงมีความสำคัญยิ่ง

This Q&A explains how client data is handled when a consultant uses cloud-based AI, and how that risk compares with enterprise employees using AI internally. It can be adopted by consultants as a client-facing guide, used by legal teams to assess AI-related contract terms, and used by executives to understand the actual exposure of company data. It also sets out the practical safeguards, governance choices and limits that should be understood before confidential information is processed.

This Q&A explains how client data is handled when a consultant uses cloud-based AI, and how that risk compares with enterprise employees using AI internally. It can be adopted by consultants as a client-facing guide, used by legal teams to assess AI-related contract terms, and used by executives to understand the actual exposure of company data. It also sets out the practical safeguards, governance choices and limits that should be understood before confidential information is processed.

Once we have captured our organisation’s knowledge, how do we actually make it usable by AI? This article aims to provide a practical mental model of some of the technical concepts involved, including retrieval, RAG, knowledge graphs and fine-tuning, so executives can better understand the options and evaluate proposals from AI integrators, whether they come from management consultancies, hyperscalers or technology and systems integrators.

Once we have captured our organisation’s knowledge, how do we actually make it usable by AI? This article aims to provide a practical mental model of some of the technical concepts involved, including retrieval, RAG, knowledge graphs and fine-tuning, so executives can better understand the options and evaluate proposals from AI integrators, whether they come from management consultancies, hyperscalers or technology and systems integrators.

If AI is supposed to learn from our organisation, where does that knowledge actually come from? This article focuses on traditional non-tech enterprises. It addresses how meetings, conversations, and everyday decisions can be captured and connected so that the organisation gradually builds a body of knowledge that AI can use.

If AI is supposed to learn from our organisation, where does that knowledge actually come from? This article focuses on traditional non-tech enterprises. It addresses how meetings, conversations, and everyday decisions can be captured and connected so that the organisation gradually builds a body of knowledge that AI can use.

How do we let employees use AI without losing control of company data, decisions and accountability? This article addresses the five guardrails for deciding what AI can access, who can use it, what it can do, and how its use should be governed and monitored.

How do we let employees use AI without losing control of company data, decisions and accountability? This article addresses the five guardrails for deciding what AI can access, who can use it, what it can do, and how its use should be governed and monitored.

AI can make your writing clearer, but it can also add habits that make it sound generic or obviously AI-written. This article shows you how to spot those habits and gives you a prompt to help keep your writing sounding like you.

AI can make your writing clearer, but it can also add habits that make it sound generic or obviously AI-written. This article shows you how to spot those habits and gives you a prompt to help keep your writing sounding like you.

Part of the Working With AI on Long-Horizon Work series, this final article turns the earlier lessons into a practical way to organise human and AI responsibilities. There is no single ideal combination of Chat, Work and coding agents. This article provides a practical framework for deciding which roles a project needs, where responsibility should pass from one to another, and what the knowledge worker should retain. The principle is simple: use the fewest AI roles necessary, but add separation when uncertainty, consequence, sensitive data or lack of expertise makes additional thinking and review valuable.

Part of the Working With AI on Long-Horizon Work series, this final article turns the earlier lessons into a practical way to organise human and AI responsibilities. There is no single ideal combination of Chat, Work and coding agents. This article provides a practical framework for deciding which roles a project needs, where responsibility should pass from one to another, and what the knowledge worker should retain. The principle is simple: use the fewest AI roles necessary, but add separation when uncertainty, consequence, sensitive data or lack of expertise makes additional thinking and review valuable.

Part of the Working With AI on Long-Horizon Work series, this third article looks beneath the interface at the environments in which AI actually operates. When AI starts editing files, running code and working with databases, it helps to know where those actions are actually taking place. Drawing on my website and Project X, this article explains local and cloud execution, GUIs, terminals, shells, sandboxes, virtual machines and localhost in plain language. The goal is not to turn a knowledge worker into a systems engineer, but to give enough understanding to know what the AI can reach and affect.

Part of the Working With AI on Long-Horizon Work series, this third article looks beneath the interface at the environments in which AI actually operates. When AI starts editing files, running code and working with databases, it helps to know where those actions are actually taking place. Drawing on my website and Project X, this article explains local and cloud execution, GUIs, terminals, shells, sandboxes, virtual machines and localhost in plain language. The goal is not to turn a knowledge worker into a systems engineer, but to give enough understanding to know what the AI can reach and affect.

Part of the Working With AI on Long-Horizon Work series, this second article focuses on maintaining continuity when a project spans many AI sessions. Chat can be a valuable thinking partner, but long-running projects quickly outgrow any single conversation. This article explains how I keep Chat aligned with the actual project through build notes, handovers, kickoff prompts and deliberate context transfers. The aim is to let Chat think with me without expecting it to become the project’s permanent memory—or exposing more sensitive information than it needs.

Part of the Working With AI on Long-Horizon Work series, this second article focuses on maintaining continuity when a project spans many AI sessions. Chat can be a valuable thinking partner, but long-running projects quickly outgrow any single conversation. This article explains how I keep Chat aligned with the actual project through build notes, handovers, kickoff prompts and deliberate context transfers. The aim is to let Chat think with me without expecting it to become the project’s permanent memory—or exposing more sensitive information than it needs.

Part of the Working With AI on Long-Horizon Work series, this first article introduces the central question of how to stay involved and in control as AI takes on more of the work. ChatGPT alongside Codex made my website build slower and more tedious, but it also forced me to stop, understand what had been built and decide what should happen next. Working more directly with Cowork on Project X showed me how easily AI execution can move faster than my own understanding. This article explores why some deliberate friction may be worth keeping when the goal is not just to build, but to remain in control.

Part of the Working With AI on Long-Horizon Work series, this first article introduces the central question of how to stay involved and in control as AI takes on more of the work. ChatGPT alongside Codex made my website build slower and more tedious, but it also forced me to stop, understand what had been built and decide what should happen next. Working more directly with Cowork on Project X showed me how easily AI execution can move faster than my own understanding. This article explores why some deliberate friction may be worth keeping when the goal is not just to build, but to remain in control.

A practical guide to building typical websites, what each one needs, and the responsibilities that add as a website becomes more capable.

A practical guide to building typical websites, what each one needs, and the responsibilities that add as a website becomes more capable.

A simple guide to the different parts behind a website, what each one does, and how they work together.

A simple guide to the different parts behind a website, what each one does, and how they work together.

Frontier AI tools are moving from chat-based assistance toward goal-driven agent work, but for complex builds, human-in-the-loop review can still be safer until the requirements, acceptance criteria, and review points are clear.

Frontier AI tools are moving from chat-based assistance toward goal-driven agent work, but for complex builds, human-in-the-loop review can still be safer until the requirements, acceptance criteria, and review points are clear.