Artificial intelligence has made consumer technology much easier to build. Small teams can now develop features, analyse information and create personalised experiences that once required more capital and specialist expertise.
As a doctor and founder of a consumer health technology company, I have become increasingly aware of how quickly credibility can be lost. A product might take years to earn trust, yet one inaccurate or overly confident response can make someone question everything else it offers.
Companies face pressure to place AI at the centre of their products and marketing. General-purpose language models can produce plausible health content and personalised responses almost instantly. But sounding informed is not the same as being clinically reliable.
Widespread AI use can therefore make trust harder to establish
Consumers may not know whether information has been clinically reviewed, generated automatically or adapted to their circumstances. Describing a product as “AI-powered” may create more questions than confidence.
Over the past several months of building in consumer health, I have had to consider where AI improves a product and where it risks weakening it. Three considerations have become particularly important.
Begin with a real problem, not an AI capability
It is easy to start with the technology. A team sees what an AI model can do and then searches for a health problem to which it might be applied. The process should run in the opposite direction.
A health information product should begin with why someone is looking for information and what they need to do next. The technology is secondary to the situation in which it will be used.
One useful question is: if the AI component disappeared tomorrow, would this problem still be worth solving? If not, the product may be relying on novelty rather than a genuine need.
Including AI in the product cannot be an afterthought, and its role should be determined by the problem. It is most valuable when it makes an experience easier, faster or more accessible, rather than becoming the experience itself.
Treat credibility as a product decision
Trust in health technology is often discussed in terms of branding
A company uses reassuring language, highlights its clinical advisers and creates an interface that looks professional.
Those things matter, but credibility is also shaped by what the product does. Why does it collect particular information? Can users understand where its outputs come from? What happens when the system is uncertain? Is generated content reviewed appropriately?
Founders must also be clear about the product’s intended purpose. There is an important difference between a general wellness product and software intended to diagnose, monitor or treat a condition, or influence a clinical decision. The latter may be a medical device and should follow the appropriate regulatory pathway, with the necessary evidence and safeguards. A disclaimer cannot resolve this after the product has been built.
Even when a product is not a medical device, its outputs can affect how someone responds to their health. Language models may sound confident when information is incomplete, creating anxiety, false reassurance or confusion about whether professional help is needed.
AI may be most useful behind the scenes, organising information or reducing repetitive work. Where it produces user-facing health information, companies must define its role carefully and keep their claims, evidence and regulatory approach aligned.
Use AI to reduce effort, not manufacture authority
Consumer health products compete not only for attention, but also for effort. People are asked to enter information, answer questions and change their behaviour. Every additional step must justify the time it demands.
I encountered this while building Moya, a tracker for people with skin conditions. Early on, it was tempting to treat every additional data point as useful. In practice, each extra question also gave someone another reason not to complete a check-in.
That changed the question from “What could we collect?” to “What would someone genuinely find useful later?”
AI can reduce repetitive entry, organise scattered information or summarise a record. These applications may be less dramatic than an AI health adviser, but they solve an immediate problem by reducing work for the user.
This requires a more thoughtful definition of engagement. Most consumer technology aims to increase sessions and time spent inside the product. In health, more attention is not always better.
A useful product might support a brief interaction, remain in the background and become valuable when someone needs to review information or prepare for an appointment. Its success is not daily attention, but usefulness at the right moment.
Founders should look beyond screen time
Did the product reduce confusion? Did it make information easier to understand? Did it help someone take an appropriate next step? These measures reveal more than anything else.
Build the product people can trust
AI will continue to become faster, cheaper and more widely available. Access to the technology itself is unlikely to remain a meaningful advantage.
The greater advantage will be judgement: knowing where AI improves a product, where stronger evidence or regulatory oversight is required, and where a simpler approach would serve the user better.
For consumer health founders, the most useful question is not, “How much AI can we add?” It is, “What does the user need, and does AI help us meet that need safely?”
Companies that answer that question honestly will be better placed to build products that people use, value and trust.
