I kind of want something to work on as a side project.
I have a really vague toy idea of "what if ChatGPT did stuff async?". So for example you have a bot that messages you at some time of the day to start a conversation. I don't know what the practical uses are, but what got me there is having a bot to motivate you towards your goals. I don't feel I have enough of a thread on this idea yet to build something, so might let this one stew.
Another idea, somewhat validated by a recent HN post, is the idea of tools to organize your life. Specifically if you own a house, there are a bunch of things you need to do every 1 month, 3 months, 6 months, annual etc. TODO lists do the job (like Microsoft Todo) which I use now but feel a bit generic. While they are good at reminding they are not so good at the "view of what is coming up". Also they lack a certain "emotional" element that I am trying to get at (which links with the bot idea), that how do you make these things feel less like chores. Something like gamification but not quite that.
Another thing that would scratch an itch for me is 24/7 hrv monitoring (using a Polar H10 or similar). Could be done via an app, or a separate device (so spare your phone battery). Then if the device has some ideas of what you are doing (maybe it asks you an hour later if you had low HRV "what were you doing an hour ago?) you can correlate HRV to daily activity. (HRV is Heart Rate Variability. See https://en.wikipedia.org/wiki/Heart_rate_variability)
So I have a few vague things stewing but not enough to even know what to code.
I also don't know if these are money ideas, or more like a Github project.
> I have a really vague toy idea of "what if ChatGPT did stuff async?". So for example you have a bot that messages you at some time of the day to start a conversation. I don't know what the practical uses are, but what got me there is having a bot to motivate you towards your goals. I don't feel I have enough of a thread on this idea yet to build something, so might let this one stew.
I saw in another thread somewhere, people talking about a simple cron job, checks maybe your email/calendar and has some algo for ranking "importance." If it's important enough, it writes you a notification.
The LLM doing the ranking of importance, basically.
I like this idea, and it would scratch my own itch - what if I forget to check my calendar in the morning but I have an 8 am meeting I have to attend? This could be done by rules, but maybe the LLM can give an natural language interface. "Alert me one hour before, for meetings booked by someone else that are before 10am"
The whole "alert me about X" but don't want alarms blaring off all day.
Typically when people are chasing ideas in quest for some investment of time they are thinking of solutions to apply instead of problems to solve. I recommend looking at problems you actually have right now. Things that consume your time, require you to look things up, or things you find frustrating and solve for that one step at a time. If it provides immediate value to you then odds are it will benefit other people as well.
Yes I know what you mean. The heart rate thing is closest to a problem to solve, although probably a problem that needs a doctor involved. I think it could be useful and would probably pay for something that did it well myself. The reason is for chronic fatigue it is an interesting metric to follow. Say I am at the gym I can be warned to sit down for 20 minutes before carrying on.
I find with tech problems I have at work, the way I want to solve it is to ditch all the damn complexity :-). Maybe the solution is a book on how to convince your org to KISS lol.
I'm working on a collection of LLM-related demos to showcase interesting use cases in the context of video applications. Things like filler word removal during recording post-processing, creating a vector index from a video library, live on-call AI "assistants", etc.
I have an idea for a product that doesn't seem to exist. I'll have to look up patents to see if it's covered already or not. Rather not say what it is.
Making ham and wifi antennas.
Trying to figure out zfs. Seems easy, until I try to add new vdevs/drives to an existing pool/vdev(encrypted zfs on Ubuntu root drive) and change it to raidz1. Would have been nice for the Ubuntu installer to allow more zfs config at install time.
Considering making extra money selling grow your own mushroom kits since the local guy in my area closed down. Doesn't seem like it's really doable without a commercial retort for sterilization, but those are insanely expensive and they don't seem to make fairly compact ones (70-100qt).
I've been working on a pdf bank statement converter. The work itself isn't inherently interesting. It's the full ownership of a thing I made, combined with actually helping people, that makes it really satisfying to work on.
I’ve been working on a small project to get familiar with building Chrome extensions. It makes tweaks to Google Flights to add data like transfer partners, estimated point redemption costs, etc. that can be helpful for folks redeeming credit card or frequent flyer miles. It also makes a bunch of UI enhancements to make the results easier to review.
Working on a write-up of the Russian ArVid system (file backup that used VHS tapes as media & your domestic VCR). Got screenshot, software, basic schematics... the next challenge is writing it up.
You can help: tell me what questions you'd like a write-up like this to answer.
Continuing to hack on https://rolepad.com, a free job application tracking app. Currently working on the employer side to allow greater visibility into application status for both parties and hopefully lead to less ghosting.
This is a cool product. From an applicant's point of view, though, it's very easy to get depressed when you get tens of rejection emails or no emails at all. When people get depressed, they stop applying, which means they stop using your product as well. Just a thought.
I'm working on yet another software transactional memory for C++. It's actually a higher-level API on top of the in-memory database I worked on at my previous startup, with many perf improvements since then (update transactions are ~200ns). I have a draft website but the API implementation isn't complete, so you can't download it yet: https://senderista.github.io/atomik-website/.
I thought of making a startup out of this project, but was talked out of it by several friends and acquaintances with startup experience. I still think it has potential as an open-source project, although I doubt it would be monetizable.
Indeed, my API is inspired by Haskell's STM: it's essentially just TVars with `atomically` and `retry` (I'm not yet convinced `orElse` is worth the implementation effort).
I’m working on a business automation platform that helps businesses quickly (within a couple of minutes) set up a fully customizable solution to run their business their way, all while being able to gain insights and forecasts easily.
I have a really vague toy idea of "what if ChatGPT did stuff async?". So for example you have a bot that messages you at some time of the day to start a conversation. I don't know what the practical uses are, but what got me there is having a bot to motivate you towards your goals. I don't feel I have enough of a thread on this idea yet to build something, so might let this one stew.
Another idea, somewhat validated by a recent HN post, is the idea of tools to organize your life. Specifically if you own a house, there are a bunch of things you need to do every 1 month, 3 months, 6 months, annual etc. TODO lists do the job (like Microsoft Todo) which I use now but feel a bit generic. While they are good at reminding they are not so good at the "view of what is coming up". Also they lack a certain "emotional" element that I am trying to get at (which links with the bot idea), that how do you make these things feel less like chores. Something like gamification but not quite that.
Another thing that would scratch an itch for me is 24/7 hrv monitoring (using a Polar H10 or similar). Could be done via an app, or a separate device (so spare your phone battery). Then if the device has some ideas of what you are doing (maybe it asks you an hour later if you had low HRV "what were you doing an hour ago?) you can correlate HRV to daily activity. (HRV is Heart Rate Variability. See https://en.wikipedia.org/wiki/Heart_rate_variability)
So I have a few vague things stewing but not enough to even know what to code.
I also don't know if these are money ideas, or more like a Github project.