A different way to find a doctor.

Most doctor directories start with the internet. We wanted to start with people we actually knew.

While building HealthFirstAI, we spent a lot of time inside hospitals and talking to doctors. We started noticing that we were building a small network of doctors without really trying to.

We had met them. We knew where they worked. We knew what they actually did.

So we built Callroom around that.

Instead of trying to list every doctor, the idea was to build a small, verified directory of doctors we had actually met.

Then make finding them easier.

A patient doesn’t always know whether they need a cardiologist, orthopedist, or neurologist. They usually just know what is bothering them.

So we built search around symptoms rather than specialties.

You could write:

“My left knee has been swollen for two weeks.”

Callroom would use semantic search to find doctors whose experience was relevant, rather than looking only for the words “knee” or “orthopedic.”

What we built

• Verified doctor and hospital profiles
• Public, SEO-friendly profile pages
• Symptom-based doctor discovery
• Semantic search using vector embeddings
• AI-enriched doctor profiles for better matching
• Hospital search built from hospital and doctor information
• Booking and contact links
• Custom domains and configurable profile themes

We also built a review product on the same infrastructure, which meant doctors, hospitals, and other local businesses could use the same underlying profile and review system.

Tech Stack

Next.js · React · Supabase · PostgreSQL · pgvector · OpenAI · Razorpay · Tailwind CSS

Built at Halfwrong, the studio I co-founded with Tejas Gurav.

Links

Website: callroom

Why we didn’t launch: Read the story →