Writing · 97 articles
Practical thinking on AI, engineering and leadership
A 2026 series on using AI in engineering companies, and earlier essays on strategy, technology leadership and my own frameworks. Follow by RSS
Latest
AI in engineering companies (2026)
Where the value is, where the risks are, and what has to be in place first. Written from hands-on work, not from the sidelines.
Start here · 13 min readAI is not my first rodeoWhat seven hype cycles taught me about this one Read itThe five levels of AI value
Find, structure, connect, act, learn: the ladder, and why you cannot skip steps.
- 1Beyond search: the five levels of AI value in an engineering company8 min
- 2Your company already knows the answer. It just can't find it.7 min
- 3From documents to data: building an engineering model you own7 min
- 4Where should your AI live? Cloud, UK-hosted or in the building11 min
- 5From what the manual says to what the product is doing6 min
- 6AI drafts, engineers decide6 min
- 7Closing the loop: letting field data change the product6 min
Agents
What agents are, when not to use one, and how to run them safely and affordably.
Security
How AI moves the attack surface, and how to defend your AI systems and your connected products.
Coding agents
Getting real value from AI coding tools: judgement, security, tokens, cost and privacy.
- 1Vibe coding is just another level of abstraction5 min
- 2Coding agents: strengths, weaknesses and opportunities5 min
- 3Coding agents need experienced engineers5 min
- 4The hobby system and the professional system7 min
- 5The security of AI-written code5 min
- 6What tokens are, and why they matter5 min
- 7Managing a token budget5 min
- 8Local, cloud and frontier: getting the most from your budget5 min
- 9Keeping your code local and private6 min
Essays
Technology leadership, connected products, strategy and the methods I use with clients: ideas I have tested in practice.
AI and technology leadership
- Write the constitution before the code
- The VenturePilot constitution, annotated
- One constitution or many? Governing AI agents across a larger organisation
- Five things I was sure of
- Autonomous internally, permissioned externally
- Why my website has an MCP server
- When AI removes the translation layer between thinking and building
- When AI decisions have physical consequences
- What a CTO should be hearing
- Not all CTOs are created equal
- Which technical debt to pay down first
- Leading without pretending to know
- What vision means in a CTO
- What it takes to be an outstanding CTO
- The C-suite a startup actually needs
- Confidence, ego and the value of listening
- Embracing the Fourth Industrial Revolution: a three-minute guide for organisations
- From silicon to society
- The AI nobody will admit to
Technology risk and due diligence
Connected products
- The CRA without a compliance department
- PSTI: the UK product security law that already applies to you
- When the product is already in the field
- Shipping is not launching
- Connected data, disconnected decisions
- The update is the strategy: OTA firmware for connected products
- Metrics that tell you the truth
- What makes an embedded system reliable
- Decoding a format you did not design
Strategy
- How to diagnose strategic deficit
- When a company has no strategy
- How much product strategy a young company needs
- Competing beyond price
- When a larger competitor dumps price
- Defined strategy vs ambiguity for pre-profit companies
- Strategy on a longer clock
- Products the market actually needs
- Revenue forecasting: strategies for uncertain times
- Strategic communication is part of the work
- The silent strategy: harnessing quietude in business innovation
- Old wisdom that still works
Organisations and people
- The older I get, the better I remember I was
- After the idea
- The way I think
- When everybody's job is nobody's job
- What downsizing does to a company
- How neurodiversity can fuel innovation at work
- Don't invest in collaboration tech until you've sorted the wetware
- Bridging the gap: collaboration across departments and organisations
- The crucial role of trained facilitation in collaboration
- The unseen beauty of inclusion at C-suite level
- A pace you can keep
- What resilience actually feels like
My methods
- Idea Mining: why good ideas need a mine, not a workshop
- From invention to Idea Mining
- Decision Mining: group decisions that actually stick
- The average said build. The odds said sell.
- Conscious AI: the same revolution, maturing
- Lean is dead: the future of work is adaptability
- Conscious Collaboration
- If you only had a hammer
- Conscious Agility
- How’s that going for you?
- When ego kills innovation: the case for Conscious Collaboration and Idea Mining
- Idea Mining and Black Swan theory: hearing the unpredictable
Archive: 3 earlier essays
Written mostly in 2024 and kept here for reference.
No article titles match that filter. Try fewer letters, or ask my writing instead.
I use AI tools in my writing, mainly to work faster and to bridge the gap between thinking and getting words on a page. The ideas, arguments, frameworks and experiences are mine, developed through direct practice over a lifetime at the bleeding edge, not generated by a prompt. Given what I advise on, it seems worth being clear about that.