My cousin is a radiologist. He built a quick prototype to automate his radiology reports. His colleagues wanted it.
That’s when he recruited me: “Can you help turn this into a real product?”
This is RadKits. Here’s what I’ve learned building healthcare SaaS with a domain expert co-founder.
How RadKits Started
Raid had a problem: writing high-quality radiology reports is detailed, time-consuming work. The templates exist, but they take forever to modify for each case.
He built a quick prototype to automate the repetitive parts. His colleagues saw it and wanted access.
That’s when he asked me to help turn it into a real product—something secure, scalable, and usable by radiologists beyond his immediate circle.
I said yes because this was different from my previous failed SaaS attempts. This time:
- The problem was validated (radiologists were asking for it)
- We had a domain expert (he knows radiology inside and out)
- There was early traction (people using his prototype)
I wasn’t guessing at a problem. The problem was real and people were already trying to solve it.
What We’re Building
RadKits helps radiologists write better reports faster by automating the boring parts.
Specifically:
- High-quality templates based on structured reporting best practices (RSNA guidelines)
- AI-assisted report writing that updates templates based on dictated findings
- Improvement suggestions from an integrated knowledge base
- Template customization (radiologists keep control over their templates)
The goal: Reduce radiologist burnout while improving report quality. Make high-throughput diagnostic work faster, easier, and less error-prone.
Where we are: MVP with users at Raid’s hospital. Planning public beta soon.
What I’ve Learned Building for Domain Experts
Building for domain experts is different from building for yourself.
In my previous SaaS attempts, I expected perfection because I was both builder and user. I’d iterate endlessly trying to get everything right before shipping.
With RadKits, I have to ship imperfect things. That’s the only way to get validation from actual users. Feedback matters more than my assumptions because I’m not the primary beneficiary.
Validating AI quality requires the domain expert in the loop.
After a decade debugging infrastructure systems where I could verify correctness myself, RadKits forced a shift: I can’t judge if an AI-generated report improvement is actually good. That’s where having Raid as both product visionary and quality validator matters—he’s technical enough to work with the AI pipelines while bringing domain expertise I don’t have.
This is the core challenge in domain-specific AI: you need the expert constantly validating outputs, not just advising on features.
Adoption patterns revealed our early adopter profile.
Younger radiologists integrated RadKits into their daily workflow immediately. Experienced radiologists used it selectively for complex cases. This wasn’t surprising—early adopters of any tool tend to be those with less investment in existing workflows—but seeing the pattern in our specific market helped us focus beta outreach on newer radiologists.
Async collaboration across 7-8 hour time zones works better than expected.
I work on RadKits in my mornings (US time). Raid works on it in his afternoons/evenings (UTC+3 time). This natural handoff has worked surprisingly well.
What’s Next
We’re preparing for public beta launch. That means:
- Finishing the beta waitlist and onboarding flow
- Incorporating feedback from our alpha users
- Setting up proper billing (navigating non-US business incorporation + payment gateways that aren’t Stripe)
The biggest open question: pricing. How much is “reducing time on reports without compromising quality” worth? We’re hoping the beta program helps us figure that out.
I’ll be writing more as we grow RadKits—beta launch, pricing decisions, and what we learn about healthcare SaaS go-to-market. Follow along if you’re interested in building with domain expert co-founders.
If you’re a radiologist interested in trying RadKits, signup for our beta.