A content designer who speaks product.
I work at the intersection of content design, product strategy and LLM-powered systems — designing customer-facing AI experiences, and the content standards that keep them trustworthy at scale. I've done content design and product side by side rather than one after the other, which is why sitting between design, product and engineering is where I'm most useful.
Ben's voice card
Direct. Curious. A little theatrical. Allergic to jargon.
I've never really done content design and product as separate chapters. I started in content design at Babbel in Berlin, then moved into product management because I wanted more influence over the experience itself — but I never stopped owning the content. Often I was the only native English speaker on the team, so the words were always going to be mine whatever my job title said.
That's the throughline: a product and content design hybrid, running both at once across Babbel, NeuroNation, what3words, Klarna and now Wise. It's why I can hold my end of a conversation with a PM about sequencing or an engineer about constraints, and why I've never accepted the version of this job where someone hands you a finished screen and asks for better words.
Today that means building customer-facing AI experiences — assistants, agents and automated flows that real customers use to sort out real money problems — and the standards, prompts and guardrails that keep them useful, understandable and honest at scale.
Half British, half Italian. I'm fluent in German and Spanish, conversational in Italian, and read Modern Languages at Durham — which is what took me to Berlin in the first place, and why localisation has followed me through every role since. You write differently once you understand that the same message needs expressing very differently from one culture to the next.
Where I lead
Four disciplines that keep showing up together, because in AI-era products they're no longer separable.
AI & LLM Systems
Prompt and context engineering, conversational agents, LLM-ready content, model output testing, and the guardrails that stop generated copy drifting.
Content Design at Scale
Voice, tone, naming, design-system guidance and content standards that let a hundred teams sound like one, without a writer in every room.
Product Partnership
I've run products as well as written for them — growth loops, experiment design, behavioural economics — so I can hold my end of a conversation with product and engineering.
Localisation & Transcreation
A languages degree and more than a decade shipping into dozens of markets. I design for transcreation, not translation — and I can prove the difference pays.
Where I've worked
The full CV has the detail.
Group Content Design Lead / Principal IC
Head of Content Design
Content Design & Product Consultant
Product & UX Manager
Product Manager & UX Content Designer
Content Intern → Senior Content Designer & PM
Selected work
Product content design at scale — AI systems, content standards, and localisation programmes built inside large organisations.
Building a prompt coach that roasts you into competence
Turning my own feedback process into a system, so 70 designers could get it at once — then watching it spread to PMs, analysts and engineers.
Making an AI assistant actually sound like Klarna
Building the tone system and content rules that made a fast AI assistant genuinely good — not just quick.
Turning payment rejections into recoveries
A cross-company content standard for the moment nobody wants to design: telling someone no.
Building a voice & tone system from nothing
A small content team, a hundred-plus product teams, and no documented voice. How we fixed that — and made it stick.
Proving transcreation beats translation
Using the app store as a testing ground to prove localisation was a product lever — and to build an experimentation culture we could take into the product itself.
Designing a voice interface people actually talk to
A content model for a product with no screen, no undo, and a failure mode that sends you to the wrong continent.
Giving users control, on purpose
What happened when we applied self-determination theory to a brain-training app's hardest feature: difficulty.
Building a certified English assessment
Two very different companies, one exam — and the particular discipline of writing content that has to measure people fairly.
Understand my perspective
Four positions I've argued for repeatedly — in writing, in talks, and with cross-functional partners who wanted to understand content design better.
A content designer's real skill isn't words. It's context and information.
Words are the visible output, which is why people think they're the job. They aren't. The actual skill is deciding what someone needs to know, in what order, at what moment, and what can be left out entirely — then defending that structure when six stakeholders want their sentence added back in. That's information architecture and judgement under pressure, and it's the same skill whether the output is a screen, an error state, a system prompt or a support flow. Hire a content designer for the words and you've bought a Ferrari to sit in traffic.
I've argued this at length →You can't partner with product and engineering unless you understand them.
This is my core strength and I'm not modest about it. I've worked as a product manager and a content designer at the same time, for years, so I can read a roadmap, argue about prioritisation, size an experiment, and understand what a constraint actually costs to change. Content designers get shut out of decisions not because people don't value language, but because they can't tell whether we understand the trade-offs. Speak the language of the room and you stop being the person who gets handed the screen at the end. You're in the conversation where the screen gets decided.
I ran a webinar on exactly this →AI should scale our impact — but only if we bring the craft to it.
The mistake isn't using AI. It's using it to go faster at the surface layer, where our value was never concentrated in the first place. The genuine opportunity is that prompting and context engineering are writing problems — defining a role, framing context, setting guardrails, specifying an output. That's a brief. We've been writing briefs for years. Designers who realise this stop being threatened by the tools and start being the people who make them work.
Prompting is designing →Transcreation beats localisation. Every time I've measured it.
Localisation usually means handing a translator your finished English and asking for it back in another language. What you get is a product that reads like a translation, because it is one. Transcreation gives local experts the same brief you used to write the English — the audience, the goal, the constraints — and lets them build their market's version as its own thing. They can throw out your headline, pick different imagery, change the colour if red means the wrong thing where they are. At what3words I ran this as a controlled experiment rather than an argument, and the numbers settled it.
See how it played out →Writing
Streams of consciousness, and the occasional hot take.
Talks & appearances
Conferences, webinars and podcasts.
Earlier & advisory work
Consultancy projects and earlier product work, for the curious.