Scaling content safely with AI
in healthcare

Designing the systems, guardrails, integrity and tooling to enable product teams to create content more independently - but without removing specialist judgement.

The challenge

As HeliosX scaled across products and markets, almost all product copy still routed through my small Content Design team. We needed to remove that bottleneck without compromising UX quality or the medical, legal and regulatory standards required in digital healthcare. 

My role

I turned an initial proposal to decentralise lower-risk copy into an operational content model. I reframed how we defined risk, brought Product, Medical, Legal and Brand together to establish guardrails and ownership, and helped translate that expertise into the knowledge and rules underpinning our AI tooling.

The solution

We created a shared knowledge base and an AI-powered Figma Content Helper that reviews UI copy in context, suggests improvements, flags potential risks and routes work requiring specialist judgement.

The outcome

The Content Helper is now being trialled by product designers and product managers, bringing content quality checks and specialist review guidance directly into their workflows.

Early feedback shows it's catching issues designers might otherwise miss, from overly complex sentences to potentially problematic claims. It's also helping teams review UX copy in context, before it reaches specialist reviewers. Beyond the plugin, we've established a review-tracking workflow and an initial Claude skill for engineering.

These early results are helping us refine the tool and understand where AI adds the most value: not simply generating more copy, but helping teams recognise when content needs greater care.

The hardest part about introducing AI wasn’t generating good copy. It was creating the conditions for people to trust it.