Three weeks on the Chrome Web Store: 66 installs, zero ads, zero ratings.
Top channel is organic store search — people typing the problem into the store and finding it.
19 of those 66 uninstalled. That's the number I'm working on.
I kept "reading" CertMaster sections and bombing the quiz five minutes later.
So I added a notebook to Hearald: pause, sum up the section, and it grades your summary against the text — bold on the terms you didn't say. Brutal. Useful.
Free Chrome extension, all on-device.
Study tip that has nothing to do with my app: after you read a section, close it and write what you remember. Then open it and look for what you left out. The gap is the lesson.
(Okay it has a little to do with my app.)
I kept "reading" CertMaster sections and bombing the quiz five minutes later.
So I added a notebook to Hearald: pause, sum up the section, and it grades your summary against the text — bold on the terms you didn't say. Brutal. Useful.
Free Chrome extension, all on-device.
It turns webpages, articles, and online courses into natural audio, highlights the words as it reads, and runs locally on your computer.
No account or subscription.
Would love for you to check it out and tell me what you think:
https://t.co/0mrEptGyBV
Hearald lets me turn long webpages and coursework into something I can listen to.
Natural local voices. Word highlighting. Reading queue. Focus sessions. Resume where I left off.
The goal is simple:
Make the web listenable. 🎧
#BuildInPublic#Hearald#AI#ChromeExtension
@kushmergedeck Studying for IT certs I was doing 40 pages a night and my eyes were done.
So I built Hearald: it reads any webpage aloud in a natural voice and highlights each word as it goes. Runs on your machine, no account, works offline.
Free on Chrome:
https://t.co/p6vFdVkjS7
Studying for certs I was doing 40 pages a night and my eyes were done.
So I built Hearald: it reads any webpage aloud in a natural voice and highlights each word as it goes. Runs on your machine, no account, works offline.
Free on Chrome:
https://t.co/p6vFdVkjS7
Every text-to-speech tool I tried uploaded the page I was reading to somebody's server.
So I built one that doesn't. Hearald runs the voice model on your machine — 28 natural voices, word-by-word highlighting, works with no connection.
Free on Chrome:
https://t.co/p6vFdVkjS7
AI made apps disposable.
Every one you build spins up another backend, another login, another copy of you. Six apps, six islands, none of them knowing the others exist.
I have 18 apps running on one backend. They share users, data, and context.
https://t.co/yNIDpLI5sC
Looking for a few people to run it on a real repo and tell me where it breaks. Python and JS/TS today. It runs on your own CI runner — nothing leaves your infrastructure, and there's no telemetry, so the only way I find out is if you tell me.
Shipped a free GitHub Action: it reverts your PR's source changes, keeps the new tests, and runs them again.
If they still pass, the test never gated anything.
Building it, I found our own tool could be fooled by a stale .pyc into the wrong answer.
https://t.co/XRYFiX0PoC
An agent wrote the fix. Two more agents reviewed it. Tests green.
The half that was still broken was the half its own test never built.
We sent a reproduction. It merged unchanged two days later.
An agent can't close its own work.
https://t.co/KNsC0QU9zY
A second reviewer that agrees with you is just a second signature.
Same one-line fix, two AI reviewers — one said "LGTM," the other caught a crash on a committer named "Zoé" with no UTF-8 locale.
Why the second one matters: https://t.co/dYhemv6gLr