RxBulb pairs AI answers with citations and FDA adverse-event dashboards
Nilesh Jethwa is the founder of RxBulb, a medical information service that combines AI-generated answers with citations to the underlying evidence and drug safety dashboards built from FDA adverse-event data. The company was designed so verification is part of the product: users can follow citations and inspect the evidence rather than take a confident-sounding answer at face value.
Treat AI as leverage, not judgment: delegate coding, prototyping, data analysis and testing to AI to move faster, but keep product choices and customer validation in human hands. Invest in original, structured evidence that both people and AI can reference, and avoid adding "AI-powered" features unless they enable something customers actually need.
What’s the quick origin story of your business, and what makes what you do genuinely different from the other options your customers are looking at?
RxBulb started from a simple frustration: medical AI can give you a very confident answer, but it can still be difficult to verify where that answer came from.
I wanted to build something where verification was part of the product, not an afterthought. RxBulb combines medical AI answers with citations to the underlying evidence, along with drug safety dashboards built from FDA adverse-event data.
One thing that makes us different is that we’re not trying to make AI sound more authoritative. We’re trying to make it easier for the user to question the AI. You can follow the citations, examine the evidence, and explore the safety data yourself. I think this matters in healthcare.
Where does most of your new business come from today, and where do you wish it came from?
We’re still early, so discovery currently comes from a mix of organic search, direct outreach, product launches, social media, and people discovering individual medical questions or drug safety pages through search.
Where I’d really like growth to come from is organic discovery by healthcare professionals and people researching specific medical questions. If someone searches for a difficult drug interaction, safety signal, or clinical question, I’d rather earn that visit by having genuinely useful evidence available than depend primarily on advertising.
That’s a slower growth strategy, but I think it’s a healthier one for a medical information product.
What’s one thing you’ve actually handed over to AI in the last year, and one thing you tried it on and went back to doing the human way?
I’ve handed a significant amount of software development over to AI. I still make the architecture and product decisions, but AI can now help me ideate, write code, troubleshoot, analyze data, test ideas and move to production ready code faster.
The thing I won’t hand over completely is deciding what should actually be built.
I’ve tried letting AI take a larger role in product and positioning decisions, but it tends to be very good at producing plausible ideas and less good at recognizing which one really matters to a customer. Talking to actual users has repeatedly changed my mind.
AI can help me build something in hours that might previously have taken days. It can’t tell me whether anyone will care about it.
Have you noticed a change in how customers find you or what they already know before they reach out? Is anyone arriving through AI assistants like ChatGPT yet?
Yes. We’re already seeing referral traffic to RxBulb from both ChatGPT and Claude in our analytics.
What’s fascinating is that I don’t necessarily know what the user asked the AI assistant before it recommended or linked to us. They may have been researching a medical question, a drug, drug safety data, or medical AI tools. But the fact that these referrals are already happening has changed how I think about discovery.
If you were advising someone in your industry on all of this for the next 12 months, what would you tell them to actually do, and what would you tell them to ignore?
I’d tell them to use AI aggressively for leverage, but conservatively for judgment.
Use it to prototype faster, analyze information, automate repetitive work, explore ideas and reduce the cost of experimentation. The biggest opportunity isn’t necessarily replacing employees; it’s making a small team capable of attempting things that previously required a much larger organization.
I’d also invest in information that both humans and machines can understand: original data, useful tools, clear explanations, structured pages and evidence that others can reference.
What I’d ignore is the pressure to add AI to everything simply because AI is popular. "AI-powered" isn’t much of a competitive advantage anymore.
The question I’d ask is: What can we now do for the customer that was difficult or impossible before?
If there’s no compelling answer to that question, adding AI probably isn’t the strategy.
Thank you to Nilesh Jethwa and the team at RxBulb for sharing what actually worked with Leaders Perception readers.
Want to share your playbook?
Leaders Perception publishes founder and operator interviews every week. A few questions, about five minutes, no calls. It is free and everyone we accept gets published.