You have identified a problem in healthcare and built an AI solution that could help solve it. Perhaps it analyses patient information, highlights patterns or supports a healthcare professional’s decisions. As you move from prototype to product, one question deserves attention early: could your software be a medical device?
For a software startup, the answer can influence much more than regulatory paperwork. It can affect the claims you make, the evidence you need, how you develop and update the product, and your plans for entering the market.
Start with what your software is intended to do
The use of AI does not, by itself, make software a medical device. A key consideration is its intended purpose: what the software is designed and presented to do, for whom, and in what setting.
Think about the difference between software that helps a clinic schedule appointments and software that analyses information to support a diagnosis. Both may use sophisticated technology, but they serve different purposes.
For founders and product teams, it helps to put the intended purpose into plain language:
- Who will use the software?
- What information will it analyse?
- What output will it provide?
- Will that output inform a decision about an individual patient?
- How will the user be expected to act on it?
If your team gives different answers to these questions, that is a useful signal to clarify the product concept before making further assumptions about its regulatory path.
Pay attention to the claims you make
Startups often refine their message as they test the market. A product might initially be described as a tool for organising information, while later website copy or a sales presentation says it can help identify a condition or guide treatment.
Those descriptions matter. Your website, pitch deck, product materials and conversations with customers should tell a consistent story about what the software is intended to do. Reviewing claims as the product evolves can help you spot when its proposed use has changed and whether your earlier regulatory assessment still holds.
Understand what classification could mean for your plans
If your software qualifies as a medical device, the next step is to determine its classification under the applicable regulation. That assessment depends on the software’s intended purpose and how its output is used. It helps establish the requirements and route to CE marking.
This is relevant when you are setting development milestones or discussing launch dates with investors and partners. A regulatory assessment made early can give your team a more realistic basis for planning, even if parts of the product are still evolving.
Plan how you will demonstrate performance
A promising prototype and positive feedback from potential users are valuable, but a medical device company also needs to demonstrate that its product performs as intended and is sufficiently safe for its intended use.
For an AI solution, practical questions may include: What data supports the product’s claims? Does that data reflect the people and setting in which the software will be used? How will the team test the software, manage risks and assess future changes?
You do not need every answer on day one. You do need a plan for developing the evidence alongside the product, rather than treating it as a task to begin shortly before launch.
Think beyond the first release
Software teams are used to improving products through regular updates. In healthcare, changes to an AI-based product may also need to be assessed for their effect on performance, safety, intended purpose and regulatory documentation.
It is useful to establish early how your team will document decisions, evaluate changes and monitor how the product performs once it is in use. This can make it easier to keep developing the software responsibly as your company grows.
Get clearer on your next steps
Every solution is different. A general checklist can help you identify questions, but it cannot determine the regulatory status of a particular product without looking at its intended purpose and how it works.
Free webinar - 29 October 2026
In our free webinar, “Is your AI software a medical device?”, Oscar He Lindholm, Client & Business Development Partner and AI expert , will discuss software qualification and classification, relevant regulations and guidance, CE marking, and considerations for quality, risk management and clinical evidence.