I built MapMates because I moved to Bangkok and realized how hard it is to find your people when you're new somewhere. LinkedIn felt too formal, dating apps weren't right, and Facebook groups are just noise.
The idea is simple: drop a pin where you are, say what you do (founder, developer, designer, etc.), and see who else is nearby. If someone looks interesting, send a connection request with a message. If they accept, you can chat.
Some technical decisions:
- No signup required. I use Supabase anonymous auth combined with a device fingerprint hash. Your pin is tied to your device. If you clear your browser, you can recover it via email (if you added one) or a recovery code.
- Location is intentionally fuzzy. Pins are snapped to ~1km precision. Close enough to find people, not precise enough to be creepy.
- Rate limited connections. 10 requests per day. Forces you to actually read profiles instead of mass-messaging.
- Built with Next.js, Supabase, and Mapbox. Deployed on Vercel. The whole thing is pretty lightweight.
The target audience is digital nomads and expats - people who move a lot and need to rebuild their network every few months. Would love feedback from anyone who's been in that situation.
Happy to answer any questions about the tech, the target audience, or why I made certain decisions. Would love feedback on what's working and what's not.
I’ve been building, launching, and documenting a bunch of SaaS tools over the last few years. After 100+ experiments, the vast majority were good but optional- meaning people tried them once, then moved on.
In reading back over what did stick and what quietly faded, I noticed a pattern:
The winners don’t just automate tasks - they automate decisions, embed into workflows, reduce cognitive load, and earn trust over time.
In my latest article, I share what I think will work in SaaS in 2026, and what I think won’t - not theory, but patterns based on real product outcomes.
Key themes you might find interesting:
- Products that decide what to do next, not just show data.
- Tools that run in the background and plug into existing workflows.
- AI used as infrastructure, not as a hype label.
- Why many standalone dashboards and single-interface outputs will struggle.
I would love to hear what HN thinks:
- Do you agree with these predictions?
- What SaaS trends do you think will matter most by 2026?
- Am I missing anything?
I built Feelr because performance tools tell us what is slow,
but not what that slowness actually feels like to a user.
Paste a URL and Feelr replays the page load as a sensory experience:
DNS, TLS, TTFB, HTML parsing, JS execution, and third-party scripts
are expressed through timing, interaction resistance, and optional
sound / haptic cues.
The goal isn’t measurement accuracy or optimization advice — it’s
intuition. After using it, you start noticing where “heavy” actually
comes from.
There’s also a simple Compare mode to feel two sites back-to-back.
No signup. No accounts. Minimal data. Just a small experiment.
I’d love feedback from people who think a lot about performance.
I didn't do anything special for Bing. My best guess: ChatGPT uses Bing for search, and my language-specific pages (Bengali, Hindi, Tamil voice-to-text) rank well there.
The tool: Free tier uses Web Speech API (audio goes to Google/Microsoft servers, not mine). Pro tier uploads files to AssemblyAI for transcription.
Nothing groundbreaking technically — just a clean UI on existing APIs. But the Bing/ChatGPT traffic pattern surprised me.
Anyone else seeing similar Bing > Google patterns?
I built MapMates because I moved to Bangkok and realized how hard it is to find your people when you're new somewhere. LinkedIn felt too formal, dating apps weren't right, and Facebook groups are just noise.
The idea is simple: drop a pin where you are, say what you do (founder, developer, designer, etc.), and see who else is nearby. If someone looks interesting, send a connection request with a message. If they accept, you can chat. Some technical decisions:
- No signup required. I use Supabase anonymous auth combined with a device fingerprint hash. Your pin is tied to your device. If you clear your browser, you can recover it via email (if you added one) or a recovery code.
- Location is intentionally fuzzy. Pins are snapped to ~1km precision. Close enough to find people, not precise enough to be creepy.
- Rate limited connections. 10 requests per day. Forces you to actually read profiles instead of mass-messaging.
- Built with Next.js, Supabase, and Mapbox. Deployed on Vercel. The whole thing is pretty lightweight.
The target audience is digital nomads and expats - people who move a lot and need to rebuild their network every few months. Would love feedback from anyone who's been in that situation.
Happy to answer any questions about the tech, the target audience, or why I made certain decisions. Would love feedback on what's working and what's not.
thank you!
reply