Up to you whether you want to limit your answer to software that is possible/realistic
I would like an app to summon the vultures when there is a dead animal in my yard. Click for vultures.
A program that continuously poisons all corporate AI data sets.
I mean, that’s sort of already happening: Model Collapse. It’s just a matter of time now.
Nice try, open source developers who want to make the world a better place.
I was thinking more along the lines of someone who wants to be an open source developer but doesn’t know how to do it so will just use AI to vibe code litterally the entire thing.
Nurse here. The software we chart with (that I’ve used at least - plenty out there I haven’t touched) ranges from meh to absolute shit. Some of the features on my wishlist would be elimination of redundant charting (like, if put the date down in one section, don’t fuckin ask in 3 others!), a way to save templates without patient-specific data, so when I do the same procedure on 5 different people, I can just click the template and fill in any holes instead of starting from scratch each time. Some actual standardization so different programs can actually talk to eachother. Like, ER, scheduling, preop, surgery, anesthesia, recovery, med surg, and sterile processing are all departments that might be involved in a single patient visit (checked into ER, sent for emergency surgery, then kept overnight for observation before discharge)… there’ll be some overlap, but scheduling, anesthesia, and sterile processing are usually each rocking their own unique software. The others are usually decently linked up nowadays, but still with the odd exception… like trying to get surgical updates from ICU so you know when to have a bed ready; but the chart is just dead and now a PACU nurse is calling report cuz the surgery finished 40 fuckin minutes ago, all because some some jackass in the ER left the patient’s chart open which locked everyone else out of signing their notes. Surprise!
…most of the issues with our charting software boils down to dumb shit like that - proprietary systems that are incapable of communicating, or just good ol’ fashioned garbage design. And it’ll stay that way, cuz oligopoly.
As someone who’s worked with data from the backend of these systems I feel your pain. Why does someone need 10 different DoB’s over the course of 3 visits and why did those 3 day visits occur on 8 different days?
As a frequent patient I understand why everyone is always asking me my DOB excessively, to make sure they’ve got the right person and that I’m ‘with it’ enough to call the shots about my health, I assume? I guess that translates to paperwork too.
Thats why they ask you yes.
They them have to fill out forms on the visit and are asked there multiple times.
most of the issues with our charting software boils down to dumb shit like that - proprietary systems that are incapable of communicating, or just good ol’ fashioned garbage design
This is everywhere. I deleted a more thought out response when I realized that someone probably handed over ALL of the backend databases of patient data to an LLM (it’s HIPAA compliant so don’t worry *wink*) so that it can spew out MORE bullshit that doesn’t work, but a lot of money was paid for it so we’re going to use it.
HIPAA is the opposite of privacy anyway - it’s specifically about enabling sharing of data 🤦🏼.
And it was sold as a privacy tool. Lying cocksuckers.
HIPAA is a framework for CONTROLLED sharing of data.
Privacy = controlled sharing of data
HIPAA has a lot of problems. But it’s not some trick.
This is such a weird take. Your healthcare professionals need to be able to share your data with specialists safely and securely. That’s what HIPAA is about from my perspective, and everyone I work with. If you want effective care, you need to be able to share your medical data without fear of it leaking. The fact that people abuse the system is a separate problem
:-(
So yet another way a normal person just trying to live life is a product for an ever increasing set of companies.
As a physical therapist, one thing that bugs me (and I know it’s not a software limitation but spending choices) is having to wedge my notes into a format designed for a physician or something else.
I got a small taste of that back in nursing school. Our care plans followed a very strict template for our clinical patients, and they were 100% geared toward med surg. My day in the OR made for an especially hard time trying to check all the boxes… I remember a whole lot of “yeah I didn’t assess for that, cuz pt was getting surgery… but, uh, here’s what I’d be looking for if I did! …uh, nope, no interventions, due to the patient still getting surgery, but here’s a few things I could have done if they weren’t being actively cut open”
Fortunately the profs were well aware of that dillemma, so they graded pretty soft on OR days lol. So, you’ve definitely got my empathy on this one - dealing with that garbage every day sounds like a nightmare!
…this is probably the core issue of why so many different departments end up using a completely different software package.
Oi,I feel you so hard on this, as a software developer who works on medical devices and software used in medical contexts, holy shit the pickings are so bad.
Why can’t someone just conjure up solutions to all of this (which is totally doable technically)? Because FDA regulation forces us to follow Byzantine release processes and meet very exacting requirements for the software end product.
I could probably knock together at least two or three of your wishlist items in a week, and it would take 1-3 years for it to make it to the first customer. Even patches and fixes I build for existing products take 3-6 months to make it out the door after they’re done.
All of that to say, these make friction points that make it difficult for developers to coordinate and standardize without introducing mountains of extra bullshit that didn’t need to actually be in the product or software in the first place
It’s always been a fool’s errand to attempt software in the education or medical space. The requirements are too onerous and haphazard to be anywhere near worth it.
I work for a division of a health tech company and I can say that we hear you and value this kind of feedback. There are folks I work with who have dedicated themselves (including myself) to seek solutions to problems like these, and as a former integrations engineer, some companies are better at interoperability than others, despite the efforts of the HITECH Act in the US. My frustration is time and getting the attention of those who greenlight projects to work on things like these. On one hand, I feel OSS could help solve these problems sooner, but on the other hand, healthcare has some very opinionated people and things could get messy.
healthcare has some very opinionated people and things could get messy.
tell me you work with clinicians without telling me.
(☞ﾟヮﾟ)☞ ☜(ﾟヮﾟ☜)
I have a random suggestion that might be a little more useful/unique than my bitching above, (which you’ve probably heard a few thousand times already…)
One area that’s lacking in EVERY program I’ve used for charting is education. We get nursing students pretty regularly, as have the previous hospitals I’ve worked at (guessing this is pretty universal - where there are nursing positions, there are nursing schools fighting over clinical slots). When I was in school, we were given access to real patient’s charts during clinicals, and we spent a lot of time doing things like sitting at the nursing station and staring at a computer monitor at lists of lab values/meds/vitals trends/etc and whipping a pen out and transposing that shit onto paper so we could take it home and write a care plan from it. Patient data could be printed, but doing so was forbidden by the school program, because HIPAA - and understandably so, given the legal ramifications for fucking that up. Even copying the data into a word doc and manually omitting names and such was still hard no for the same reason - we might miss something and accidentally violate HIPAA.
So, the idea this is all leading up to: in the ‘View’ tab or something, make an option for “student view” that pulls up a plain text summary of the patient – but without any PHI. Anything entered into a patient’s name field will output as “PATIENT”; staff entered will just be “DOCTOR, DOCTOR2, NURSE, TECH” etc. Patient’s age? Round to the nearest 5. Dates? Fudge all of them by a random number of days (RNG on generating a report) , but by a consistent amount for all dates in that report, ie real admission date of Dec 3 2025 and discharge date of Dec 5 2025 --> RNG applies -42 days --> Student View summary reports Admission at Oct 22 2025, and discharge at Oct 24 2025. Labs? Fudge it by RNG with +/- 1% of real value.
You get the gist.
Give me a report that’s useless as fuck for the real patient - no PHI at all. BUT, you’ve essentially just generated a completely fake patient with useful data for a case study. In the student’s mind, there’s no difference - “PATIENT, 75, MALE, hospitalized for 2 days” is effectively the same case study as “Leeroy Jenkins, 73, Male, hospitalized on these specific dates, 2 days apart”. Slightly tweaked labs? who gives a shit, all they care about is if they’re out of range by a clinically significant margin.
And here’s a golden bit: PRINT THAT SUMBICH!! Or fuck it, email the .txt to your student email so you can copy and paste it. There’s no HIPAA, so it doesn’t matter. What matters is the student isn’t wasting their clinical hours fucking around on a computer - get their ass into a patient room where they can do some nursing, cuz they’ll learn more there than the entire rest of the degree program. They can crunch the data for a care plan at home.
…that ended up being really wordy. Sorry about the wall of text. On the off chance anything comes of it, let me know!
So there’s no software to do what literally every insurance swears they don’t break the law doing en masse?
There’s none that I’m aware of that basically anonymizes patient data for use outside of healthcare.
Billing and insurance all falls under HIPAA - those facets and every other aren’t disallowed (legally) from things like printing sensitive info - basically if it’s being used as part of your care, it’s fine.
And students are no exception - when they’re on a clinical rotation, they’re on that patient’s healthcare team and can legally access and print etc that info. The problem is that students still need that data for use outside of that care, and HIPAA doesn’t give a fuck about their homework, so any papers that have PHI on it become a violation when they start being accessed for those things. And schools don’t trust students not to fuck that up, so they’ll come up with overkill-by-design policies like disallowing students from printing PHI, even if it’s for use on that clinical and meant to be shredded before leaving. So, the school wants to play things extra safe.
…this isn’t to say that insurance companies aren’t abusing your data, but can legally handle it within the scope of your billing.
An immortal worm that drains wealth from billionaires, launders it, and redistributes it as some kind of deus ex machina bonus to everyone else or to charities.
A software version of taxes!
I was going to add that if the government(s) do it we could call it “taxes”… but decided that brevity was more important :-)
Public service announcement that, once split up, this would be a significant but not life-changing amount of money. At least in the West.
gotta rely on ex-wives of billionaires instead i guess
Okay so you know how Ad Naseum is a thing? It’s like uBlock but it clicks on ads to fuck with advertisers?
I wish we had something like that for youtuber sponsors built on top of SponsorBlock. Like imagine it finds the affiliate link, signs up for this shit using a burner email, and then the bot doing this deletes itself.
It makes the youtuber look bad, it makes the brand look bad, and you feel like a gigachad for doing it.
Is it possible? Maybe, but realistically I believe this is a form of fraud, so I imagine no one would do this, BUT I WISH IT DID EXIST!
You had me in the first half, but very quickly I realized we went opposite directions with that idea.
Like, I would absolutely love a program that automatically followed affiliate links for whitelisted channels that I watch. Obviously I wouldn’t want it to buy things, but it would allow me to show a little bit more support.
Just replied to a comment above but yeh I’d love something like this. Something that punishes or profits (in my favour) from pushy advertising.
This should not be fraud. It’s simple resistance.
I want a MMORPG where there are 100,000 unique NPCs all run by AI with a custom backstory that keep track of the world events and respond accordingly.
I want them to have player/party specific memories and information. I want the merchants to remember the player and treat them according to how they were treated.
Just go outside, man.
How do I quick save? I hit F5, shot a dude in the face, hit F9, and nothing happened…
You have to have two traits for your character from the start: white skin and incredibly wealthy. If you don’t have both of those you’ll have to become a police officer which is easy enough, but easier if you’re char is white and rich.
But outside, nearly nobody gets what they deserve.
And if you need a quest, go and find my frying pan!
…which in the end you’ll discover a group of robber that associated with a syndicate that involve scam center, money laundering, sex trafficking, and also kidnapping. Go forth, my hero, for my heirloom frying pan needs your help.
I need ten frying pan handles, so you must keep hunting frying pans down in the Skillet Sepulchre until you’ve got ten handles. Frying pans only have a 10% chance of dropping handles as loot. Frying pans have a five minute respawn timer. There will be a dozen other players trying to complete this quest.
Khajit has wares, if you have coin.
Side quest: go grind from 9:00 to 6:00 pm, 5 days a week (minimum), till you get enough coins.
Yes, that is something that modern AI could have some value in a game. Not AI-generated assets (graphics and music), not even AI filling in for lazy writters, but AI providing emergent story and gameplay. Like Dwarf Fortress, but LLMs instead of some “simple” hardcoded rules. This should be a genre of its own, not something pushed to every game, though.
It’s a great idea in theory, but the reality is that LLMs just aren’t actually very good at that kind of thing.
Depending on the temperature of the model their writing is either painfully bland and repetitive, or it’s the kind of “creative” that’s driven by randomness, so you’ll get completely out of pocket replies that just don’t fit the situation at all.
The answer to what this person is looking for is simple; find a good group to play tabletop RPGs with. Software is never going to give you that experience.
I think it depends on the harness.
Like the LLMs are not good at keeping narrative structure intact, but with a harness that could create a playbook of known information for each AI character that the current LLM that you are interacting with can reference, then it could contribute to emergent gameplay and help build a relatively vibrant world.
The real downside is that it would require an entire data center as a bare minimum to run all of the LLMs, like the technology is still too raw and new for it to make financial sense unless people were willing to pay $500 or $600 a month to play this game.
It obviously isn’t exactly what you’ve described but you should check out Erenshor
That already exists, it is called social media bot farm.
Masturbator/Dildo software that integrates with an EEG to scan your brain, know when you’re about to cum, and then edge you over and over again.
Too difficult. Just gonna build a machine that gently zaps your brain’s cum center continuously.
You know we have invented drugs more powerful that that? However they kill you sooner or later.
No thanks, satan. My body is a temple. /s
I think it is fair to assume all drugs that give you feelings compared to orgasm are addictive as hell which makes them very easy to overdose :)
Mankind would be doomed if such a drug existed. Well maybe not doomed but like you’d notice a severe drop in birth rates.
Have you heard of https://maustec.io/pages/edge-o-matic
Something that would remove clickbait articles, AI slop, propaganda, commercials and probably more that are poisoning our lives.
Please take my money. I would pay thousands for this
Let me quickly throw the task at the LLM of my choice and start a campaign to promote it. /s
There goes 90% of the internet …
Sounds wonderful!
Sturgeon’s law. Ninety percent of anything is crap.
Honestly, existing alternatives and frontends get surprisingly close.
A multimedia centre like Kodi or Jellyfin with the ability to schedule shows like it was a regular TV channel. With shows playing on their time slots, a random movie starting every day at ten.
There are times that I just want to turn on the TV and watch whatever is on without having to scroll through the entire library for something I could be in the mood for.
have you looked at ersatztv?
Seconding this. ErsatzTV is what you want.
No, I guess I will spend the next few hours looking into it.
FYI, the dev has abandoned this software. It shouldn’t matter too much, because you should NOT put it on the internet as it has not auth at all. So it shouldn’t matter too much. But just fyi
Thanks for the heads up.
I always get nervous with these warnings because I never use anything remotely so I don’t really know what it means to put it on the internet. Can it be put accidentally or it requires a series of steps for it to be accessed?
Depends on your setup. Is it at home behind NAT? You’ll be fine. If it is on a vps, you’ll need to block the port on your server.
when was it abandoned? The git repo seems to be reasonably active
Oh. Maybe the dev came back to it. Last I heard (about maybe a year ago) the dev had stepped back from it. Good to know that it’s back being updated though!
edit: ah nice. it’s being developed again. just checked out the github repo. cool!
Multiple things exist to do this.
Jellyfin has this awesome option “sort library randomly”. I use that for music and it is great for discovery of music or to find music I haven’t listened to in years. So maybe try this for movies and watch the first one?
I use kodi on a pretty basic level for about 15 years now. I’m almost sure that somethi g like this should exist, no?
Optical character recognition/digitisation thingy but for sheet music.
This apparently does exist and is called OMR(Optical Music Recognition) software
It exists, but every bit of software I’ve tried has incredibly poor performance on anything except the absolute cleanest typset printed pieces.
I wonder if an AI can do that? It’s obsoleted LeafSnap.
just tried with the free version of claude. It works a lot better than anything else I’ve tried but it’s not perfect, averaging ~2 mistakes per 10 bars for a relatively simple score.
Seems strange that it’s so hard to get right. I guess it’s a bit like reading handwriting - subtle tiny differences?
Does it give you a midi file from a scanned musical piece?
It outputs ABC notation, not midi. MIDI isn’t really the right format for sheet music.
Wikipedia has a section on how it works and at least to some degree also why it’s hard: https://en.wikipedia.org/wiki/Optical_music_recognition#Approaches_to_OMR
If I was to build something like this, could you beta test? Sounds like a fun weekend project for me 😄
yeah, I’d be happy to test it.
Shoot me a DM and send me some sheets to work with, I’ll post a repo link with a proof of concept and we can refine it from there
This is essentially what I was about to say. I want to be able to scan my sheet music and end up with a perfectly clean, crisp, beautiful copy of it on my tablet.
Should be doable with current AI
An easy one I would do myself if I had more time. A calorie tracker that does not try to get you into a cult.
- The only time it accesses the internet is to update its on-board database.
- None of your data gets sent anywhere unless you choose to share it (for things like new foods).
- No suggestions.
- No exercise programs or tracking
- No sleep tracking.
- Tracks only what you want. Extras available through add-ons or something, but I don’t care about tracking nutrients, fat, carbs, etc. so keep them out of my way.
- In short, follow the Unix guideline of an app doing one thing and doing it well.
If you run Android, check out “Energize” on fdroid.
Got it installed. I’ll try it tomorrow.
Waistline kind of works like that.
And I think there are a few self-hostable solutions out there.
If all you need is calories couldn’t you just use any counter app for that? At that point you’re just adding numbers in.
The key is that I don’t want it connecting to a remote site every dang time I try to make an entry. Once a day for db updates should be all it needs.
Accurate search engine that strips ai generated articles from search results.
Accurate product search engine, that finds what I am looking for instead of ads, and includes searching by what the product is made of, it’s shape etc. Last time I was trying to find no glue no glass non-welded no plastic tea infuser it took me far too long.
I’ve used https://kagi.com/ for the past year or so. It’s much improved search compared to current Google.
Literally just got pissed off enough to give up for a bit searching for a hard drive enclosure. Fucking pages and pages of docks. I typed in the god damn parameters of what I am looking for, stop showing me shit that isn’t that. The assholes miscategorizing their shit so it comes up in more searches should be executed.
Working on the second: https://menagerie.circuitforge.tech/snipe
Still early days but the filters and trust scoring work really well
An actual, healthy, good online social ecosystem. A place with deep quality conversation without bots, trolls, rage bait, misinformation, etc. Somewhere that felt like actual community.
Sadly that’s a human problem that technology is hard-pressed to solve (if not impossible).
I hear Lemmy is pretty good but I’ve never used it.
I am afraid that would only be possible if it’s faked, with everyone else except you being a LLM
Or/and if almost every real user is shadow banned to other real users except bots
Real talk I’m surprised some tech bro hasn’t made that yet. Like not hiding it, just fully branding it as an AI-first replacement for Reddit/Twitter.
The user could describe the type of content they want to see or “world event” they want their “community” to experience. Like a medival twitter, and have a political plot develop between two kingdoms and the human user’s engagement on the thread can help sway public opinion on the subject leading to different narrative events.
With a world of 8 billion people, there are enough out there that would enjoy it…
Nice try, Microsoft. Not letting you vibe code my ideas
