Now

What's on the table this week.

A "now" page. What I'm reading, thinking, building, watching. Snapshot, not a blog post. Updated when there's something worth saying.

Last updated: 10 August 2026

Building

The Modern SEO + AI Optimisation Standard is out of the workshop and into friends-testing: ten pillars, 27 checks, a scorecard you fill in as you read. It's the document I kept wishing I could hand new clients on day one, so I wrote it. It joins the little bench of free tools next to the audit, the schema sniffer and the ask volume checker.

And the AI-visibility work I mentioned last time is now a weekly automated run: four engines (ChatGPT, Claude, Gemini, Perplexity), live web search on, every answer judged for mention, rank and sentiment, then diffed against last week's. The diffs are where the actual insight lives, which is a point I keep making to clients who want a single score.

Most recent piece: Google's AI Mode cites Dog on the Table. Until I ask a second question..

Reading

A short, biased list of what I've been saving lately. It refreshes itself from my reading each week.

Thinking about

Why a single AI-visibility score is a category error. I made the long version of the argument in The 14% problem, and running my own monitor every week keeps sharpening the short version: the number moves for reasons that have nothing to do with the brand, and the week-to-week churn is the finding, not the noise.

Also: the EU's AI-disclosure rules, live as of this week. Most coverage treats them as a compliance chore. The more interesting question is what routine labelling does to trust. Once "made with AI" is stamped on half the content in the feed, the value shifts to whoever can credibly sign the other half.

And, still, idly: how completely markdown has won. Plain .md text files are now the format the big models most like to be fed, and the one they reach for when they answer. There are probably more .md files in the world right now than there ever were WordPerfect documents, and WordPerfect once ran every office on earth. The dullest format in the building, casually eating the lot.

Where I'm spending too much time

The play section. This is a marketing consultancy site, and yet it now hosts PENTA, a daily five-round game that somehow earned itself a full v2.0 with an outline round and a practice mode, and Dog on the Decks, a generative nu-disco groovebox with a bark button. No strategy document anywhere says these should exist. Building them is the most fun I've had all month.

And chasing the new models, same as ever. OpenAI shipped Luna, Terra and Sol, and I've been running all three through my agentic workflows to see which of last month's assumptions still hold, with Kimi K3, the current open-weight darling, along for comparison. Equal parts research and procrastination, and I know which side is winning.

Open offer

If you've got a brand and you're wondering what AI actually says about you right now, I'll run a selection of tailored, customer-relevant prompts through ChatGPT, Claude, Gemini and Perplexity and send you back the raw output, no commitment. It's quick. It's free. It's a strong indicator of whether we should talk.

Take me up on the offer →