What Buyer Questions Should a Medical Device Manufacturer Test in ChatGPT, Gemini and Claude?

Who asks, and what do they want to know?

What questions should you test? A starter bank

Replace [category] with your device type, such as infusion pumps, patient monitors or surgical staplers.

Shortlists - "What are the leading [category] manufacturers for hospitals?" - "Best [category] for ambulatory surgery centres?" - "Which [category] do large health systems standardise on?"

Regulatory status - "Which [category] devices are FDA 510(k) cleared for [indication]?" - "Is [product] CE marked under the EU MDR?" - "What is [product] approved or cleared for?"

Clinical evidence - "Clinical studies comparing [product] with [competitor product]?" - "Evidence that [category] reduces [complication]?"

Interoperability and cybersecurity - "Which [category] integrate with Epic or other EHRs?" - "[Category] with HL7 or FHIR support?" - "How does [product] handle cybersecurity and software updates?"

Cost, contracts and service - "Total cost of ownership of [category] systems?" - "Is [manufacturer] on a group purchasing contract for [category]?" - "Which [category] manufacturers offer on-site service within 24 hours?"

Safety and recalls - "Has [product] been recalled?" - "Known safety issues with [category]?"

Where to buy - "Where can I buy [product] in [country]?" - "Authorised distributors of [manufacturer] in [region]?"

Brand - "What does [manufacturer] make?" - "Is [manufacturer] a reliable supplier?"

How should you run the test in ChatGPT, Gemini and Claude?

1. Use each assistant with web search available, since device facts change.
2. Run each question three times in fresh conversations.
3. Record manufacturers and products named, whether you appear, the facts stated about your products, and the sources cited.
4. Check facts against your regulatory records: clearance numbers, indications, CE status, recalls and field safety notices.
5. Repeat quarterly, and after any clearance, label change, recall or major publication.

The three draw on different search systems. Gemini draws on Google Search and Claude's web search uses Brave. ChatGPT uses third-party search providers and its own index, and studies disagree on how closely its citations follow Bing's results. A product can be well described in one and missing or outdated in another. Choosing which assistants to monitor is covered in which LLMs brands should monitor.

What sources do assistants cite for medical devices?

• Regulator databases, such as the FDA's 510(k), recall and adverse event databases in the US, and notified body or EUDAMED information in the EU.
• Peer-reviewed literature and clinical trial registries.
• Independent evaluators and research firms for health technology.
• Manufacturer product pages, instructions for use and technical specifications.
• Distributor catalogues and group purchasing announcements.
• Trade publications covering hospital technology and procurement.

If assistants state facts about your products from old sources, make sure current information is published clearly on your own site and in the public records you control.

What promotional rules should you keep in mind?

Device promotion is regulated. In the US, promotional claims should stay within a device's cleared or approved indications. In the EU, the MDR prohibits misleading claims about a device's intended purpose, safety and performance. When you publish content to improve how AI describes your products:

• keep product pages to cleared or approved indications, with appropriate safety information;
• have regulatory and legal teams review pages and press materials, as with any promotion;
• do not try to steer AI answers towards claims you could not make yourself.

If an assistant states an off-label use or wrong indication for your product, correct the underlying sources you control and record the issue. Do not publish content that amplifies it. In the US, FDA's 2024 draft guidance on addressing misinformation about medical devices and prescription drugs describes how a company can voluntarily correct third-party misinformation with tailored responsive communications; have your regulatory team check whether it covers the sources behind the AI answer before responding.

What should you fix first?

1. Accuracy of regulatory facts: indications, clearances, CE status and recall status on your site and product pages.
2. Product pages with specifications, interoperability and service details in plain text.
3. Evidence pages summarising published studies with links.
4. Distributor and contract information that tells buyers where and how to buy.
5. Presence in trade and independent sources that assistants cite for your category.

What do people ask about testing AI for medical devices?

Should we test patient questions for professional devices?

Only if patients influence the choice. For home-use devices, yes; for hospital capital equipment, focus on professional buyers.

How do we handle a wrong recall statement in an AI answer?

Check the regulator's current records and your own recall pages are clear, then correct outdated third-party pages where possible. Record the issue for regulatory and quality teams.

Do assistants recommend specific manufacturers for clinical questions?

Sometimes, but they are cautious with clinical advice. Procurement and specification questions are more likely to name manufacturers.

How many questions are enough?

Around 40 across the groups above, run three times in each assistant, is enough for a baseline.