About John Bowman

Who is John Bowman?

John Bowman is an AI strategist and developer based in Leeds, UK. I've spent more than 20 years in iGaming and digital marketing, and I now design, build and ship AI tools under my own name at JohnB.io.

Term: AI strategist. Definition: someone who works out where AI belongs in a business and where it doesn't, then proves it by building the thing rather than presenting a slide about it.

The career underneath that is unglamorous and specific. Multilingual poker and casino operations across 40 countries. Fraud investigations. Over £150k recovered in chargebacks. Live event media. Then digital and SEO leadership across 24 language markets, and product ownership of cross-functional teams above 90 people with £1.5M+ of budget attached.

What do I build at JohnB.io?

Free tools, mostly. More than 20 of them are live and none ask you to sign up: the T&Cs Simplifier, the iGaming Compliance Map, the LLMs.txt Generator, an AI Copyeditor, and a free 35-lesson AI and Machine Learning course with a certificate at the end.

Every one is a static HTML page plus a Cloudflare Worker that holds the API key and the rate limit. No framework, no build step, no package manager. That's a deliberate choice and I'd defend it: a page that loads in under a second and still works when JavaScript fails is worth more than the same tool wrapped in 400kb of React.

That principle got tested on this page. The skills panels above used to be generated by JavaScript, which meant every AI crawler that reads the site without running scripts saw six unlabelled buttons and nothing else. They're rendered in the HTML now.

Why does 20 years in iGaming matter for AI work?

Because iGaming is one of the few industries where a bad automated decision comes with a regulator attached. Send the wrong bonus message to a self-excluded customer and it isn't a bug, it's a fine.

So the questions I ask about any AI deployment come in a fixed order: what happens when the model gets it wrong, who signs it off, and what the audit trail looks like. That's why I built the AI Governance Framework and the compliance map before I built anything more interesting.

The failure I see most often is a company buying an LLM licence and calling it a strategy. Most of the value sits in the dull part around the model: who reviews the output, what gets logged, which decisions a human still has to make. Nobody wants to hear that in a kick-off meeting.

What did 88 million AI music views actually pay?

£600. Between 2024 and early 2026 I made 60 AI-generated tracks, distributed them through SoundON Global, and they picked up 88 million views. Nine of them passed a million each. The total payout was six hundred pounds.

I keep coming back to that number because it's the cleanest lesson in AI economics I've had first hand, and it complicates the point I made earlier about building things yourself. Generating output is now nearly free. Distribution isn't, and reach and revenue stopped being the same measurement somewhere around 2024. The full write-up is here: what happens when your AI songs get 88 million views.

Where did John Bowman train in AI?

The University of Oxford AI Programme in 2024 is the qualification that matters most, followed by IBM Generative AI Engineering in 2026. Before either of those: a Professional Diploma from the Digital Marketing Institute in 2016, New Technologies for Business Leaders at Rutgers University in 2020, and a run of CalArts UI/UX courses through 2021.

More recent additions are DataCamp's Responsible AI Practices and Understanding the EU AI Act, both completed in 2026. The full dated list sits in the Education panel above.

Certificates don't make anyone an engineer, and I'd say that with a long list of them sitting right above this paragraph. They're a record of finishing things. Useful, and not the same as being able to ship.

What do I write about?

Practical AI: how it behaves rather than how a vendor deck says it behaves. Recent pieces cover generative engine optimisation, what genuinely changed in AI models over the last 12 months, and the free 35-lesson AI and machine learning course that explains the terminology properly for people tired of nodding along.

I don't publish to a schedule. Pieces go up when I've got something I'd want to read myself, which means three in a month and then nothing until August.

How do you get in touch?

The contact form is the fastest route, and it's also where UK GDPR data requests go. LinkedIn works if you'd rather. Code lives on GitHub. Replies usually take a few days.