Writing
Notes from
the work.
What I learn building and marketing software for law firms, agencies and startups. No theory I have not shipped.
What is WebMCP? A practical guide for people who run websitesWebMCP lets a web page hand AI agents a list of actions they can call, instead of clicking through the page. What it is, where it stands, and what to do now.
AI intake for solicitors: a working patternHow a law firm can use AI to triage new enquiries safely: what the system does, what it must never do, and where the solicitor stays in charge.
Jamstack, headless, and the return of the serverA look back at 2021, when static sites and headless CMSs were the answer to everything — what they got right, and why the server came back.
Remote work, and the clients who had never used a video callA look back at 2020, when every client meeting moved online overnight — what changed in how projects ran, and the habits worth keeping.
The year everything became a single-page appA look back at 2019, when brochure sites were rebuilt as JavaScript applications — what it cost in speed and search, and when an SPA is still right.
From Elance to consulting: what changed and what didn’tLooking back from 2026 at a decade of client work, from Elance bids in 2013 to direct consulting — what the platforms taught me and what I stopped doing.
RAG, agents, and the approval step nobody designsA look back at 2024, when retrieval and agents went into real business workflows — and why the review screen mattered more than the model.
What a law firm actually needs from its websiteWorking with UK solicitors taught me that a law firm website has a narrow job: build trust, meet the SRA rules, and make it easy to get in touch.
Core Web Vitals, and measuring what users actually feelA look back at Core Web Vitals: what the three metrics measure, why field data beat lab scores, and what small sites should do about them now.
Page speed became a ranking factor, and everyone panickedA look back at 2018, when Google made mobile page speed a ranking signal — what clients feared, what actually moved, and what was worth fixing.
The Angular 2 rewrite, and what it cost the teams who followedA look back at 2016, when Angular replaced itself and a lot of working software was rewritten for reasons that had nothing to do with the people using it.
Babel, ES6 and the years we all had a build stepA look back at 2016, when transpiling stopped being exotic and every project grew a build — and at what we gained and lost by it.
React arrived and jQuery didn’t dieA look back at 2015–2017, when React won the argument about how to build interfaces — and most of the web quietly carried on with jQuery anyway.
The first year of AI in client work: what actually stuckIn 2023 every client asked about AI. Looking back, the things that survived contact with real work were narrower — and duller — than the demos suggested.
WordPress paid the bills. Here’s what it actually taught meYears of WordPress work for small firms taught me more about marketing, trust and maintenance than any framework has — including when not to use it.
Node in 2015: what we believed, and what actually lastedLooking back at Node.js in 2015 — callbacks, the io.js split, and which of that year's certainties are still in my projects a decade later.Prefer this in your inbox?
Say hello and I will send new posts as they go up.