MAnoj

We launched Synapse. [ Read till end ] And instead of posting another “we’re changing healthcare with AI” thread, I want to share the questions people have been asking us. Doctors. Hospital owners. Founders. VCs. Customers. Some of them were uncomfortable. Probably the most useful questions we’ve received. “Why would Apollo need you if OpenAI or Microsoft can give them models, servers, engineers and basically unlimited compute?” Fair. We are not trying to build a better GPT. And we’re not trying to replace Apollo’s AI stack. We’re building the layer underneath it. Healthcare doesn't happen inside a prompt box. It happens while a doctor is walking through a ward, examining a patient, speaking to an attendant, ordering a test or deciding what happens next. Today most AI systems wait for that information to reach a computer. Synapse goes to where the information is created. A small wearable on the doctor. It listens with consent, understands the clinical conversation and turns it into structured actions. workflows. Notes. Investigations. Referrals. Follow-ups. Appointments. Patient instructions. EMR updates. Eventually, one conversation should be enough. Doctor says: “Get an ultrasound tomorrow and come back Friday at 4.” Synapse should understand that. Create the order. Schedule the follow-up and call. Send the patient instructions. Update the clinical record. No typing. No opening five different apps. Someone asked: “Why not just use the doctor's phone?” Because possible ≠ usable. Try asking a doctor seeing 80 patients to unlock a phone, open an app, position it correctly and manage recordings 80 times. The interface has to disappear. Another question: “So you're a hardware company?” Not exactly. Hardware is our entry point. The pendant captures the real world. Software structures it. Integrations make it useful. Clinical memory makes it compound. The long-term product isn't a pendant. It's clinical intelligence infrastructure on top of data. Then came: “What if OpenAI launches the same hardware?” They probably can. So can Apple. Microsoft. Google. 100 Shenzhen manufacturers. The moat cannot just be a plastic pendant with microphones. The moat has to become: hardware clinical workflows hospital integrations deployment doctor behaviour structured clinical data intelligence of hospitals distribution If oneday thousands of doctors across hundreds of hospitals use Synapse every day, replacing us shouldn't mean replacing a device. It should mean replacing infrastructure. Another doctor asked: “AIG/Apollo is already working on similar things. Isn't that competition spedning over 4-15cr per month?” I actually see it as validation. If the biggest hospital groups are spending serious money building AI infrastructure, the question is no longer: “Will hospitals use AI?” It's: “What does the final stack look like?” And I think hospitals will use many models. OpenAI today. Microsoft tomorrow. Their own model somewhere else. Synapse should work underneath all of them. One line I've started using: We don't want to replace GPT inside the hospital. We want to become the device that gives GPT clinical context. Then VCs asked the question every founder eventually gets: “What traction actually matters?” Not how beautiful the demo is. Not how many AI agents we can show. Devices deployed. Doctors actively using them. Consultations captured. Hours of documentation saved. Pilot → paid conversion. Revenue per doctor. Retention. That's it. So our GTM is intentionally boring. We're starting with smaller private hospitals. 20–30 beds. 5–7 doctors. Not because the opportunity is small. Because that's where we can break things, learn quickly and fix them. Large hospitals give you one clean shot. Small hospitals give you iteration. Small hospitals are where we learn. Large hospitals are where we scale what we've learned. Another interesting conversation was around cameras. Yes, eventually Synapse can have vision. But just because hardware can have a camera doesn't mean healthcare needs one everywhere. Privacy matters. Consent matters. Trust matters. Audio is enough to solve an enormous problem first. Vision should only appear when it creates genuine clinical value. Then came the biggest vision question: “Where does this go if it actually works?” Today: Give every doctor a second memory. Next: Give every hospital a real-time clinical intelligence layer. Eventually: Imagine thousands of hospitals producing privacy-preserving, structured signals in real time. An unusual symptom suddenly increasing across one district. A medication pattern changing. A cluster appearing weeks before traditional reporting catches it. The same infrastructure helping one doctor write a note could eventually help healthcare systems understand what is happening across populations. That's much further out. We have to earn the right to build it. Right now the mission is much simpler: Build something doctors actually wear. Put it into real hospitals. Watch where it fails. Fix it. Repeat. We have the software running. Hardware is moving into deployment. The next batches are being built. And we're spending more time with doctors than inside pitch decks. Building Synapse has already changed one thing for me: The best questions about a startup rarely come from startup Twitter. They come from the customer sitting across the table asking, “Why do I actually need this?” That's the question we're trying to answer every day. wearsynapse.com