How Six Years of Helping People Inspired a Multilingual AI Platform and a New Generation Social Network
At a time when most social networks compete primarily for user attention and artificial intelligence is rapidly becoming part of everyday life, one project is attempting to combine both technologies with a single goal: helping people.
Most monitoring tools tell you when your site is down. PulseBoard tells you why.
I built PulseBoard because I was tired of getting alerts at 3 AM and spending hours digging through logs, CloudWatch, and GitHub commits just to figure out what went wrong.
Here's how it works:
Monitors your websites and APIs Vigil AI investigates incidents automatically Correlates failures with GitHub commits and deployments Pulls AWS telemetry (EC2, RDS, ALB, Lambda, CloudWatch) and now Lightsail! Scores commits by time, file relevance, deployment status, and infrastructure correlation Filters out noise below a 40/100 threshold to save AI tokens Shows you the evidence, not just a guess
I made a feature [ AI Response Cache Chatbot ] of [ airespaker ] AI Response Taker. Users can take AI response from system's cache, user's cache or directly from AI (Google Gemma). Users can take URIs of displaying AI response or screenshot of it. By tapping 'C' button, URI of displaying AI response is copied to clipboard. By tapping 'S' button, URI of screenshot of AI response is copied to clipboard.
I'm testing Haku, a small Telegram-native AI companion beta. The product thesis is simple: most companion chats feel good on day one, then start to feel generic when corrections, tone, boundaries, and running context don't survive across days.
The beta is focused on memory + continuity rather than a bigger character marketplace. It is early, free to try, and I am mainly looking for people who care enough to stress-test whether the companion still feels coherent after several conversations.
My kid asked to watch one Bluey episode. Auto-play had other plans. Forty-five minutes later it was unboxing videos, a bedtime meltdown, and me wondering why a video platform built for "watching what you want" had quietly turned into a platform that decides for you.
That's the problem I set out to fix with NoSuggest, a free, open-source PWA that strips YouTube down to just the channels you choose. No recommendations, no Shorts, no auto-play, no trending feed, no algorithmic rabbit holes.
Kids only ever see content from channels a parent explicitly added. There's a dedicated Kids Mode with a 6-digit PIN, and it hides anything that could lead to an accidental tap outside curated content.
I've been trying to analyse my own dreams for a while but every tool out there gives the same generic answers with zero context about your actual life.
We built compress.aurorapdf.com because we kept running into the same problem.
Every time we needed to shrink a scanned contract or a heavy report to squeeze it under a 25MB email limit, the only options meant uploading the file to a third-party cloud server. For sensitive work documents, that never felt right to us.
Same app you use to run models locally. You pick a model from the catalog, select a GPU, and deploy. Edge checks if the model fits the hardware before anything spins up, same way it does for local. No failed deployments, no wasted GPU hours.
Your cloud deployments sit next to your local ones in the same interface. One place to track costs, monitor usage, and swap GPUs.
