My AI Automation App Got Bunch of Stars and Downloads on Github
I posted about Tasker (https://github.com/pitalco/tasker) on Hacker News 2 days ago and to my surprise it got a bunch of downloads and stars on Github.
I built Tasker because I was looking for an AI automation application very specifically built for people like my father who is a self-employed HVAC technician. I wanted to help him automate his estimate workflows (you would be SHOCKED that this is the majority of time spent for self-employed HVAC technicians).
I built it as an open-source desktop app (cause thats just what I wanted), slapped a random name on it (yes its a generic name, I know and there are other apps named Tasker) and started using it. I used it for a few weeks for my own sales outreach for other work while he used it for his estimates. It works surprisingly well. I shared it and was shocked by the response.
Curious if others find it useful and if anyone has suggestions for next steps. One request which is a great one is adding more "guardrails" around the AI. Have been thinking of the design for that but its a great suggestion!
Nice milestone — stars and downloads are a great early interest signal, especially for developer tools.
From a dev perspective though, I’ve found the real inflection point is when users come back without prompting — e.g., repeat installs, issues/PRs from real usage, or people wiring the tool into their own workflows.
Curious — beyond stars/downloads, what behavior signal has been most useful for you in telling that people are actually relying on the automation (not just trying it once)?
Thanks for the comment! I was not intending to pursue this in anyway like that initially TBH so I am not really tracking any of the behavior signals I prob should. But I would say workflow runs prob would be the best instance of what you are mentioning showing non prompted intent.
That makes sense — and honestly, that’s very common early on.
You’re spot on about workflow runs. In my experience, unprompted re-runs (especially after a day or two) are one of the cleanest signals that an automation has crossed from “interesting” to “useful.”
Even lightweight tracking like:
tends to surface real dependency faster than stars ever will.
Sounds like you’re already intuitively looking at the right signal.
Hi Joe
I checked your website, it looks so good.
I am a senior developer with 8+ years experiences.
my portfolio https://leebin.gamer.gd/
I'd love to contribute to your team