The problem Pingl /proof of irl connection solves

We're living through a loneliness epidemic, despite being constantly connected 24/7 and using social media to connect with friends. Screens have replaced presence, and no one is rewarding us for putting our phone down.
Pingl changes that.
Wear it. Connect. Earn on-chain rewards.
Pingl is a wearable that recognizes the moments that feel the most meaningful and warm to us : real life conversations filled with shared laughter, and good vibes, and mints uniquely generated on-chain rewards for every meaningful interaction you have in the real world.
No audio is ever recorded or stored. Pingl listens only for the patterns of positive interaction: the cadence of friendly conversation, the sound of laughter. It's fully privacy-first, processing sound signals without capturing speech content, so you stay present instead of glued to a screen:, rewardable proof of irl connections.

Challenges we ran into

There were many challenges, one of them was not having the right name for my device embedded in the bluetooth connectivity logic, and that took a lot of trial and error to get right to get the bluetooth pairing to work. The device name embedded in the connectivity logic had to match exactly what the browser was scanning for, including spaces and case sensitivity and when it didn't, the device simply wouldn't appear. It sounds like a small detail but it took a lot debugging, requiring multiple iterations of uploading new arduino code, testing, and adjusting before the connection was stable.

Another hard problem was defining what a "conversation" actually is. It's easy to detect sound, but knowing when a chat ends required us to think carefully about silence, how long a pause is just a breath versus the actual end of an interaction. This meant building in silence-detection logic alongside the sound detection, then carefully tuning the threshold so that brief quiet moments wouldn't prematurely close an active session. Getting that balance right was important to get meaningful, accurate data rather than fragmented or inflated interaction counts.We had to tweak the threshold for sound data to measure silence effectively in order to accurately measure when a conversation ended.

Use of AI tools and agents

We used AI tools such as claude, cursor and gemini to do things like help write the logic for some of the functionality, creating the digital art used for the nfts, and debugging problems.