HHS released a massive dataset of every Medicaid payment to every provider in the US: 227 million rows covering $1.09 trillion in spending across 617,000 billing providers. The data was released explicitly to crowdsource fraud detection.
The raw data is a 2.9 GB Parquet file. I built MedicaidSpending.org to make it searchable and browsable.
You can search by provider name or NPI, browse by state/city/specialty, and see individual provider pages with monthly spending trends, billing code breakdowns, and automated billing flags for statistical outliers.
Some of the patterns are striking. Brooklyn alone accounts for $31.8 billion in personal care services (code T1019) _ more than most states spend on all Medicaid combined. Some authorized officials control hundreds of billing entities. Early analysts scanning just 0.16% of providers flagged $90 billion in likely fraudulent payments.
Technical details:
- Go single binary, ~15 MB
- 3.3 GB SQLite database (read-only, pre-aggregated from the 227M rows using DuckDB)
- 900,000+ indexable pages generated from 13 templates
- No JavaScript framework _ server-rendered HTML, Chart.js for one chart per provider page
- Runs on a single VPS behind Caddy
Over the weekend, I built FeelBetterbot (https://feelbetterbot.com/), a therapy chatbot powered by AI. I posted it here earlier today and didn’t get any interest and was feeling a bit down, so I had a chat about it with FeelBetterBot. They suggested that maybe FeelBetterBot wasn’t a good fit for the HN audience, so I created CofounderBot, your virtual cofounder, powered by AI.
I’m sure that story sounds a little precious, but I’m finding real value in talking with these sorts of bots. Behind the screens, they’re just a thin wrapper around ChatGPT, but a bit of tuning for a specific use case seems to make the whole interaction much more natural. Please try it out and let me know what you think!
FeelBetterBot is a therapy chatbot using the ChatGPT API. It’s designed to help you talk through any issues you might be having and will suggest techniques from modern evidence-based therapy modalities if appropriate.
I had the idea because I’m a bit of a psychology nerd and, while I haven’t done a ton of therapy myself, I’ve got a lot of out reading people like Albert Ellis and Carl Rogers and using some of the techniques from modalities like CBT. I’ve also seen lots of people close to me struggling with mental health and finding and keeping a therapist seems to be a big part of that. Therapy can be great, but it seems to be hard to find someone you really connect with and limiting your interactions to a single, specific hour each week can leave you unsupported when you most need it. So the idea of a confidential and anonymous therapy-like chatbot, available any time you need it, seemed like it might have some value.
I’m just now diving into the world of therapy chatbots, but it doesn’t seem like there’s anything similar that’s as open and accessible as this. The closest I can think of is 7Cups, that’s that’s real people not AI. (Am a huge fan of 7Cups, though, and have been a listener there for years.)
Technically, it’s just a very thin wrapper on the ChatGPT API. Other than the interface, all I really supply is the prompt. The app itself runs entirely on Cloudflare Pages and the frontend is just Alpine.js and Simple.css. I’m a Rails guy, but Cloudflare Pages and Cloudflare Workers have become by go-to for rapid prototyping like this because it’s such so quick and easy.
Hope some of it you find it useful, and I’d love to know what y’all think!
FeelBetterBot is a therapy chatbot using the ChatGPT API. It’s designed to help you talk through any issues you might be having and will suggest techniques from modern evidence-based therapy modalities if appropriate.
I had the idea because I’m a bit of a psychology nerd and, while I haven’t done a ton of therapy myself, I’ve got a lot of out reading people like Albert Ellis and Carl Rogers and using some of the techniques from modalities like CBT. I’ve also seen lots of people close to me struggling with mental health and finding and keeping a therapist seems to be a big part of that. Therapy can be great, but it seems to be hard to find someone you really connect with and limiting your interactions to a single, specific hour each week can leave you unsupported when you most need it. So the idea of a confidential and anonymous therapy-like chatbot, available any time you need it, seemed like it might have some value.
I’m just now diving into the world of therapy chatbots, but it doesn’t seem like there’s anything similar that’s as open and accessible as this. The closest I can think of is 7Cups, that’s that’s real people not AI. (Am a huge fan of 7Cups, though, and have been a listener there for years.)
Technically, it’s just a very thin wrapper on the ChatGPT API. Other than the interface, all I really supply is the prompt. The app itself runs entirely on Cloudflare Pages and the frontend is just Alpine.js and Simple.css. I’m a Rails guy, but Cloudflare Pages and Cloudflare Workers have become by go-to for rapid prototyping like this because it’s such so quick and easy.
Hope some of it you find it useful, and I’d love to know what y’all think!
i came across this post looking for an ai therapist. I'm surprised why there aren't more attempts on this. Therapists are pretty expensive and everyone is going thru some kind of mental health crisis nowadays. With all the tech layoffs happening a lot of people, like me, are looking at cost effective therapy options.
What else are you planning on doing to it to improve it?
You sound super interesting and I’m sure plenty of people would be interested in chatting (including me!), so please think about putting your email in your profile!
I have personal experience working with pimartin. If you're looking for a reference, they really know what they're talking about for SOC 2. They helped me get SOC 2 at the company I co-founded, ProcedureFlow where I'm VP of Engineering.
My main concern was this: we are a growing company and I didn't want to bolt on "some corporate SOC 2 thing" just to make us seem more secure. Honestly, my attitude was similar to the OP.
Sales were getting blocked and delayed by lack of SOC 2 but also having to fill out security questionnaires for every customer. I found pimartin and they really showed us how SOC 2 is customizable and isn't black and white like most people think. SOC 2 is not prescriptive about how you do things. He also helped us find an auditor that understands our business and made the process very easy for us.
When our prospective customers now do their IT/Security reviews, we pass with flying colors because of the changes that have been made to our organization and the big attitude shift we had about it. SOC 2 is not a burden in our company.
Happy to talk more about our experience with pimartin and doing SOC 2 "right"!
I’m not sure how the title got changed. But “Much philanthropy is a routinized exchange between salaried bureaucrats” is the caption on the header image not the actual title.
I would imagine that Dang changed it because of the tendency of articles with baiting questions for headlines tend to result in people skipping reading over the article and going right to answering the baiting question.
The end of this article raises the issue of whether Fly.io’s USP, deploying app servers close to your users, is useful for run of the mill web apps. And as much as I like Fly.io and the people associated with it, I’ve wondered this myself. It just seems like serving to US customers from any major US data center is generally fast enough. And I think this might even be true for the world of HTML-over-the-wire web stuff, which Fly.io seems to investing heavily in.
No doubts there are plenty of more niche uses (if I were serving users internationally, I’d probably use Fly.io), but the use case just doesn’t seem as broad as the Heroku/PaaS comparisons make it out to be.
Not only that, having one location for a world-wide user base is usually enough. You can optimize much more through rendering speed, blocking requests etc than by being closer to your user.
And even if your page becomes really popular, 3 locations (Europe, US, East Asia) are enough to be <200ms to any user in the world. And it keeps your setup and cost much lower.
I guess I count as a “low wage worker” these days and your claims in the first paragraph simply aren’t true for myself or most of the other low wage workers I know.
I deliver for DoorDash and make around $100 dollars a day from about five hours of work. My girlfriend (who’s disabled) rides along with me, so it’s sorta like I’m just getting paid to drive around and hang out. I drive a 2005 Honda Civic, so I spend less than $15/day on gas and the car is old enough the depreciation is minuscule. I buy my health insurance on the exchange. It’s excellent and with the premium subsidy I only pay $22/month. I have several chronic health conditions, so good health insurance (and particularly drug coverage) is really important to me. I’ve been able to get my insurance to cover all of the many medications I’m on (many of still under patent and not generic) and my copay is only $45/month. Visits to my primary care doctor are free and I’ve got a $500 deductible for everything else.
I can work as much or as little as I want on pretty much whatever schedule I choose. I tend to stick to about five hours a day because it doesn’t feel like a ton of work, but it’s enough to cover my expenses and then some. Rent is my only other major expense and I pay just $550 for my one bedroom apartment. I rarely cook at home, but even after spending way too much eating out, I regularly have $500-$1,000 leftover at the end of the month, so I’m certainly not living paycheck to paycheck. Since I save some of that extra money, I can take time off whenever I want (although I rarely do, since I enjoy getting out of the house and working). Sure, if I get sick and don’t work, I don’t make any money, but that’s why I keep a bit of money in savings.
It might be Uber or Lyft or Instacart instead of DoorDash, but I know several people in a similar position to me. And honestly it’s a pretty sweet life. I’ve got all my needs met and then some. I enjoy my work. And I’ve got lots of free time to purpose my intellectual interests like coding.
That was my situation. I had a sales job the was primarily commission based and I had health insurance. I had a bad bike accident and got sent to an out of network hospital. I went from a six figure income and no medical bills to almost no income and a six figure medical bill. I was devastated.
Probably the hardest part was how depressed I was, even as my body was recovering, I simply wasn’t prepared to start hustling to get my sales income back up, pay for physical therapy, deal with the hospital & insurance company, etc.
For-profit healthcare should be criminalised. It's hard enough to recover physically from an injury, but adding the need to recover financially from it as well is cruel.
The problem is not for-profit healthcare, but how it is implemented in the US. Lots of countries have for-profit healthcare without the users of that healthcare every worrying about recovering financially.
At the end of the day, it's always "for profit". Either it's the doctors getting paid by the state getting paid by citizens, or the hospital getting paid by the state getting paid by citizens, or the hospitals getting paid by the insurers getting paid by citizens, or very often a combination of the 3.
The issue in the US is the setup, not the fundamental principal
Getting paid for your labor isn't the same as "for profit". Hospitals can get their stuff paid for without there being "profit", and you don't have to expect sick folks to have extra money to pay for middle men.
> Hospitals can get their stuff paid for without there being "profit"
They can, but it's not necessary. There are lots of private hospitals in France and it works just fine
> you don't have to expect sick folks to have extra money to pay for middle men
There are always going to be middlemen. Too few and resources are allocated inefficiently due to a lack of management. Too many and it's a waste of resources. The trick is to find the right balance
"There are always going to be middlemen. Too few and resources are allocated inefficiently due to a lack of management. Too many and it's a waste of resources. The trick is to find the right balance"
Yes, agreed, though I'd argue that middlemen aren't actually needed - as actual middlemen are folks getting paid when they aren't really necessary. It is kind of like car dealerships don't really need to be as they are, but that is part of the service you pay for with cars. Instead, you simply need enough people to get things to work - and sometimes that can be lessened. For example, standardized medical billing and coding in the US would reduce the labor required to deal with insurance - and suddenly, fewer people are needed at the office. Fairly standarized coverage would help as well (fewer misunderstandings) and decoupling it from employment (US based: Would take away HR jobs, but reduce burden on companies).
But the main point really was that sick people, who could easily be missing work to see the doctor, shouldn't be expected to pay for everything.
I imagine most contract delivery workers have considered this and factor it into their decision making. I don't think these positions are meant to be a perfect improvement across the board to traditional starting wage jobs, but it's up to the employee to make the judgement based on their own risk tolerance.
Like many folks here, I worked full-time minimum wage for years before eventually entering the professional world, and I can say given the choice now, I'd absolutely prefer the make-your-own-hours contract model to obligation of a fixed-hour in-person position. Of course there are drawbacks, like the lack of health coverage (though I'm Canadian so I'm not sure how this would work in the states exactly), but given the option, I'd still choose the former.
> If a car smashed into me and I couldn't code, my company wouldn't pay me either?
Mine would (not a company, a university). For the first two weeks, I'd use accrued Sick Time then switch to their Short-Term Disability that would cover 100% of my pay (for some positions with fewer years of service it would be 75%). Short-Term Disability coverage lasts for six months, beyond that my employer doesn't pay any salary replacement but I pay a few dollars a month for long-term disability insurance that would.
I think the federal Family and Medical Leave Act only guarantees I can return to my job for 12 weeks but Massachusetts increases that to 20 weeks. By policy, my employer won't dismiss someone on Short-Term Disability (26 weeks) and may not beyond that either.
In this case, driving a car is part of his job, so indeed it would be like having an industrial accident - your employer is going to cover you for a work accident
These gig job workers are classified as contractors, not employees (except where a judge or a local law says they must be employees). I don't know how, or if, workers' comp applies to contractors, I suspect it doesn't apply at all.
Let me know how your gig economy retirement plan goes. May you never have an accident on job. May you never need healthcare. I don’t see much future in such gig economy jobs. Our grandparents often worked for the same company all their lives. Your gig economy is a one way trap with no exit or promotional path. You have the equivalent of a fancy summer college kid job, but it’s not a career path.
Imagine a world where people were provided free health care, a benefits system, good employee rights and state pensions! Oh wait, it already exists in most of Europe, Canada, Australia and New Zealand
Did you read the comment? They have healthcare and said there's $500-$1000 leftover each month, which is a healthy amount for retirement investment. More than enough to max out a Roth IRA. That's on top of social security too.
And you appear to have no dependent, and living in an area where cost of livings are rather low according to your rent cost.
I'm not implying we should all afford a living in San Francisco, but an anecdote that represents the average case would be more relevant. The prospect of having a family with a couple of kids to take care of is quite the norm (still)
Really interesting to hear some insight it from someone who is actually doing it for a change. Not having lived in the US for a while I wasn't aware that insurance was available for this rate with pre-existing conditions. Sounds a lot more reasonable than what i've heard - do you have more information about your insurance plan?
Theirs does sound like a good plan (won't find that in every state) but they're also relatively poor and receiving a large subsidy to cover most of the monthly premium.
Thanks for providing some factual real-world insight to the topic of US healthcare where most information that bounces around Europeans is along the lines of "yeah, well Americans have higher salaries because healthcare costs are so extreme so everyone without a tech job is living paycheck to paycheck and can easily be bankrupted by illnesses".
Can you explain how your central kitchen fits into this? Because you don’t mention any automation there. Are enough of labor costs concentrated in the in-restaurant prep such that it’s economical to automate that while still doing the central kitchen work with actual people?
The raw data is a 2.9 GB Parquet file. I built MedicaidSpending.org to make it searchable and browsable.
You can search by provider name or NPI, browse by state/city/specialty, and see individual provider pages with monthly spending trends, billing code breakdowns, and automated billing flags for statistical outliers.
Some of the patterns are striking. Brooklyn alone accounts for $31.8 billion in personal care services (code T1019) _ more than most states spend on all Medicaid combined. Some authorized officials control hundreds of billing entities. Early analysts scanning just 0.16% of providers flagged $90 billion in likely fraudulent payments.
Technical details: - Go single binary, ~15 MB - 3.3 GB SQLite database (read-only, pre-aggregated from the 227M rows using DuckDB) - 900,000+ indexable pages generated from 13 templates - No JavaScript framework _ server-rendered HTML, Chart.js for one chart per provider page - Runs on a single VPS behind Caddy
Data sources: HHS Medicaid Provider Spending dataset, NPPES provider registry, HCPCS code descriptions, OIG exclusion list, NUCC taxonomy codes.
All public data, no login required.