Operating · accepting briefs Work with me

Get new articles by email. No spam, unsubscribe anytime.

Articles

Write-ups from the work. MCP servers, AI workflows, content engineering, search, and the occasional tangent. Most of these come from problems I hit on real engagements, written up so the next person does not have to figure it out from scratch.

What's agentic AI?

67 articles · showing 25-48

How to Use Claude Code: From a PRD to a Deployed App

In today's post I'm going through how to use Claude Code properly, from the PRD I write by hand to a deployed app, with the real repos, tests, git flow and mistakes from the things I've built with it.

claude-code · mcp
Best Local LLM for Coding, Plus the API and Image Models I Use

In today's post I'm taking a closer look at the best LLM for coding, both local and over an API. Here are the local and API models in my setup, the job each one does and the tests behind it, plus the image models I use.

claude-code · local-llm
How to set up vLLM in Docker on Windows (WSL2): serve an open-weight model on your own GPU

vLLM doesn't run on Windows natively, so this is the Docker Desktop and WSL2 route my rig has served on since July: the compose file, the image pin that stops it upgrading itself, the walls WSL2 puts in your way, and how the settings were found, with Claude writing the test harness.

local-llm · vllm
How to run houtini-lm on vLLM

I moved my local models off my old server and onto vLLM in Docker. In today's post we're taking a closer look at pointing houtini-lm at it, and why thinking goes off for a model served under an alias.

claude-code · mcp
How to do an SEO audit with Claude

SEO Audit Console is a free, open-source MCP server that pulls your crawl and your Search Console data into a database on your own PC, and Claude runs the audit from it. In today's guide we're taking a closer look at what an SEO audit covers and how the fixes get ranked and written, run for real on houtini.com with no Semrush or Screaming Frog.

mcp · claude-desktop
Best GPU for Local LLMs in 2026: What I'd Buy and Why

In today's post I'm going through the best GPU for running LLMs locally, and why memory bandwidth and VRAM decide it. I start with the cheapest VRAM you can buy and the cheapest 32GB cards, then the modded 48GB RTX 4090s I run, and finish on when an RTX PRO 6000 or a DGX Spark makes more sense.

local-llm · gpu
Claude Desktop System Requirements (2026): Minimum & Recommended Specs

Claude Desktop was my introduction to AI and it's still the best on-ramp to useful AI at work, so in today's guide I'm taking the system-requirements question the way I'd answer it for a client: the official minimums, the July change that moved Cowork into the cloud, what still runs on your machine, and why the laptop you own is probably fine.

mcp · system-requirements
Claude Code System Requirements (2026): Specs, Setup, Gotchas

In today's guide, we're setting up Claude Code on Mac, Windows and Linux in the forms it now ships in: the terminal CLI, the Code tab in the Claude Desktop app, and the VS Code and Cursor extension. The published spec, the install routes Anthropic recommends, Fable 5.1 on every paid plan, and why the machine matters less than the way you work.

claude-code · cli
AI for the CFO: How to Get AI Spend Under Control

In today's post I'm looking at why AI spend now moves with how much the tools get used, which is how Uber went through its 2026 AI budget by April, and at the four levers a finance team can pull to bring it back under control.

strategy · claude-code
How to prevent AI hallucinations in your work: the fact-check I run, and what the research says

A single factual error on your site and you're toast. Here's the check I run: my own MCPs find what needs checking and get a second opinion from a separate model, and the primary page settles it. I ran it for real on Saturn's moon count, a number from this page that was already out of date, and set it against what the 2025-26 research says reduces hallucinations.

Claude · strategy
What is RAG, and what could it do in your company?

Your AI confidently answers questions about your business with public-internet knowledge. RAG is the architecture that gets your actual contracts, customer list and operating playbook into the conversation, with citations. Here is what it is, where it sits next to long context and MCP, and what it changes if you sponsor the work this quarter.

Claude · strategy
A beginner's guide to Claude hooks

Claude Code hooks are small scripts that fire automatically at specific moments in a coding session. They give you deterministic control where CLAUDE.md instructions only get probabilistic compliance. This beginner's guide covers what hooks do, the five events you need to know, the exit-code gotcha that catches almost everyone, and the community projects worth installing before you write your own.

Claude · claude-code
What Are AI Agents? Agentic AI Explained in Plain English

You can write a good prompt and maybe run an MCP or two, but nothing happens unless you start it. In today's post we're taking a closer look at what an AI agent is, how agentic AI differs from the chat you already use, and the first agent to build in Claude Code.
Claude Code API Key Security with 1Password: A Guide to Token Hygiene

The simplest possible setup that keeps your production tokens out of AI chat windows. 1Password CLI, op run, and the conversational discipline that makes the rest of it work.
AI for the CEO and Managing Director: How to Implement It

A lot of the work in your business still runs through someone's hands, and that's where AI pays. In today's post I'm looking at why it has to be led from the top, and how I'd organise a small team to do it, with the 2026 evidence alongside.

Claude
Implementing WebMCP on a Recruitment Website

Thinking about what, exactly, the future of a website "looks" like in the agentic era is a challenging proposition. It might be that in most cases, our future viewers/readers/customers can do everything, from their chatbot of preference…
How to Set Up a Claude Code Project in 2026 (And What Goes Where)

In today's guide we're walking through the anatomy of a proper Claude Code project - CLAUDE.md, rules, skills, agents, hooks, MCP servers and plugins - and the five-step setup I use for every new repo.
Swapping the Engine: How to Run Claude Code on Local Silicon for Zero Pennies

Claude Code's real power isn't the Anthropic model sitting behind it, it's the agentic loop: the file-system access, the tool use, the way it chains tasks together without you babysitting every step. I figured this out the expensive way. I ran…
A Beginner's Guide to Claude Computer Use

I've been letting Claude control my mouse and keyboard on and off to test this feature for a little while, and the honest answer is that it's simultaneously the most impressive and most frustrating AI feature I've used. It can navigate…
DGX Spark Alternatives and AI Mini PCs: Do You Actually Need One?

NVIDIA's DGX Spark and a wave of tiny AI mini PCs promise local LLMs in a gold box. I've run local models on everything from water-cooled 3090s to a dual-4090 vLLM rig, so here's my read: the real DGX Spark alternatives, the one spec that decides it, and whether you need any of them.
How to make SVGs and images in Claude with Gemini MCP

Gemini MCP gives Claude Google's Nano Banana Pro image model, Veo video and SVG generation, with the results shown in the chat. In today's guide we're taking a closer look at where it came from, how to set it up in Claude Desktop and Claude Code, and what it made: a photoreal desk, a printed label that came back letter for letter, and the three showcase pieces from Paper's Quiver launch, traced to SVG from Nano Banana Pro images.
Yet Another Memory MCP? That's Not the Memory You're Looking For

I was considering building my own memory system for Claude Code after some early, failed affairs with memory MCPs. In therapy we're encouraged to think about how we think. A discussion about metacognition in a completely unrelated world…
The Best MCPs for SEO and Content Marketing (Research, Publish, Measure)

I ran an SEO agency for sixteen years, so I know exactly where the billable hours go - and it's not the clever part. In today's post I'm sharing the content-marketing and SEO MCP stack I now run through Claude, organised by the phase of work where each one earns its keep, including seo-audit, the one I built to kill the grunt.
How to set up LM Studio: GPU offload, developer mode and your first local model

Turn a gaming GPU into a working local AI model with LM Studio - free, no billing, no API key. In today's guide, we install it, point it at the GPU, download a first model, and then go through the settings: GPU offload, context length, Flash Attention, temperature and top-k, and the Developer mode that unlocks them.

claude-code · local-llm