Ask HN: What problem in your industry is a potential startup?
This has been done two other times before:
https://news.ycombinator.com/item?id=13139638 (2016), https://news.ycombinator.com/item?id=9799007 (2015)
Both threads generated a lot of really interesting discussion, and I was curious what the discussion would sound like if this were asked again in 2018.
252 comments
[ 3.6 ms ] story [ 227 ms ] threadFrom a business side of things: - loyalty program/rewards - pricing optimization - financial services to underserved customers including entrepreneurs, families, millennials, freelancers, etc.
High level programming language to program robots with safety in mind would be amazing. RoboDK is the only one that is doing this and they still suck.
Of course, installations are usually expected to function for ten to twenty years at a minimum...an order of magnitude greater than anything your typical js-framework-of-the-day considers.
What I want is a PLC: 1) A PLC which has a hard real-time process and soft real-time processes. Beckhoff and B&R do this, other PLC's do this as well. 2) I want to do the hard real-time programming in Rust or Ada, or "Safe/restricted C"... Rust and Ada are so much more expressive than structured text. 3) I would also like to have a simple API to allow deterministic, hard real-time communication between the real-time control domain and a soft real-time domain so that higher level languages can be used for control problems. And I would like the PLC to support one of the high level languages: CLISP or Racket, F#, Julia, Python, Elixir... don't care what it is. This is actually doable on Beckhoff but the tight connection between hard and soft real-time was not really there.. and beckhoff runs windows. Would prefer Linux, VXworks, or QNX.
I want to write code and deploy, not spend most of my time on ops or marry myself to a cloud vendor. I started my company as the only tech person and I feel like I have to be more admin than dev even though I'm doing the same thing as everyone else.
i like it though, could really help in some situations.
In software there are no such standards, or rather there are too many standards, and new ones are being released every day. Currently there is no clear incentive for vendors to standardise, and it's quite possible there will never be, it's early to tell.
What you're describing currently requires enormous resources, and no single company can pull that off -- even Google is struggling with keeping both back-end stability and front-end simplicity (I'm currently working on a new GCP tool which just got into beta, so I can see it firsthand). Standardisation allows many different vendors to focus only on a part of the whole supply chain, while with the cloud you have single vendors trying to cover everything under one roof.
It would be great if we could combine Digital Ocean's droplets with Google's BigQuery with Amazon's RDS, all through a simple front-end UI -- but the nozzle size is not standardised, and I wonder if it will ever be...
0 - https://docs.openshift.com/enterprise/3.1/install_config/agg...
You are describing a large-scale setup boilerplate, far away from what I'd call a startup in a box. Slap in a simple server like my own project [1] or no server at all until product-market fit, and then but only then start scaling.
[1] https://serverjs.io/
I've experienced many of those tools in the context of a 200-person company; I can definitely see how managed versions of them could deliver value to a much smaller org.
It seems to me that you are introducing technical debt from day one because it’s cool.
I've found this is only kinda true. Most people need the same supporting infrastructure and most people only need to deploy their HTTP-visible app, maybe a cron or backend daemon, and probably a data store. Everything else from logs to metrics to alerting to HA gateways and so on are quite universal.
But ultimately all these arguments converge to: something that beats k8s's suckiness is a potential startup.
One problem is that it requires a minimum of 3 instances, so it can be more expensive than Heroku to start.
[1] https://convox.com
[2] https://github.com/convox/rack
The Application Runtime (CFAR) is like an OSS Heroku, installable on AWS, GCP, Azure, vSphere, and OpenStack. Then just push your app and scale it out. It was first built as an enterprise Heroku competitor, and the commercial flavors power a bunch of big companies’ infrastructures. It even uses the same “buildpack” model that Heroku uses to provide language and framework dependencies.
The Container Runtime (CFCR) packages k8s in a way to make it easier to deploy, maintain, and scale, and it also deploys and manages the health of the underlying VMs. It originated in collaboration with Google.
Source: I’m a Cloud Foundry Foundation project lead.
I'm assuming you want the product to be built with open source components, so then the startup would be selling a glue script to package these open components together.
The only feasible way to make money with this would be to sell a proprietary glue script. Not sure how acceptable that would be to you or to potential customer base of this startup.
So to rope in consulting offers, you'd be inclined to add complexity and config options to generate more avenues for consulting.
In my experience, the companies that succeed are those who are sincere in wanting to help their customers.
If find PaaS can be a little too opinionated. You get boxed-in too quickly. Whereas IaaS is too raw. I had a bit of a head scratch when I recently moved to Google Cloud and was a bit flummoxed on where to put my (backend) session state. Similar experience trying to get Blue-Green deployments on GKE (Kubernetes).
I feel GCP/AWS and friends will get there. But for now there are definitely some gaps.
The point is to find problems to solve not just ideas.
I did the original posting of that question and wrote an essay about it afterwards:
http://000fff.org/the-problem-with-problems/
It is hard to compile a stack - which would be useful for most companies.
Yes, there have been attempts made - for Example, Open Stack. But then I feel, there is always an element of specific technology based on the needs of the product being built.
Robin enables an App-store like user experience to simplify deployment and lifecycle management of Big Data, NoSQL and Database deployments, On-Premise and in-Cloud. It supports application-level snapshots, clones, time-travel, QoS, scaling, backup/restore, etc. It takes care of HA, stable hostnames, Application templates, events, notifications.
Check out the demo clips for [video]
Cloudera https://vimeo.com/213037162/a66e0b4e77 HortonWorks https://vimeo.com/227832200/8d9c749984 MongoDB https://vimeo.com/206348836/a36e535add ELK Stack https://vimeo.com/225899966/6000bae6f2 Controlling IOPs in shared environment https://vimeo.com/171608156 1-Click deploy, snapshot, the clone of entire Oracle https://vimeo.com/195549219 Even RAC cluster https://vimeo.com/223730427/e8cb8c92f8 and SAP HANA many more applications
Disclosure: I am the lead developer at Robinsystems
We directly interface with those interested in the results of the check, and there is an overwhelming amount of work in building integrations with schools, applicant tracking systems, hospitals, public records, courts...
We spend most of our time building XML and JSON parsers to cram their data into our models.
If there was a company that provided a single interface to this data, you could write your own ticket. I know we aren't the only company in this space with this issue.
VoIP, especially multi-tenant voip like conferencing is 70% dependent on YOUR local network connection and 30% the provider's infrastructure. Having a high bandwidth pipe is no guarantee for 100% clear and jitterless voip communication. Have you had a network engineer take a deep inspection of voip traffic to see if there's any QoS or packet filtering going on that could degrade performance (i.e. if you're a small office that has your voice traffic occurring side by side with literally every other network device you're GOING to have a bad time).
You can join a concall via phone number + pin, from your desktop, or from the mobile app. The mobile app will even let you view screenshares or video conf.
[0]https://aws.amazon.com/chime/
https://meet.jit.si
The dedicated app runs on just about everything'; iOS, OSX, Windows, Android.
On phones you can use Video and Audio or just plain Audio (Low Bandwidth mode).
You can send your BlueJeans room URL and they can join with just a web browser (Web browsers get less features though). Non-Hosts can join the meeting without signing up for anything. They just click and join.
I think the audio and video is pretty good. Much better than Facetime, which is not a good comparison because Facetime is pretty terrible. Audio quality is good.
There is dedicated hardware that supports BlueJeans so you can use it in conferences rooms.
You can record meetings.
All the standard stuff like screen share and what not. You can even ping pong back and forth. I worked with someone where they had to type in some stuff and I had to type in other stuff on their computer. They just joined my BlueJeans room and when I was not typing, they could type. No clicking to take control or give it back, it was more or less seamless.
Possibly something that mixes business and process models into it, but again, something simple where you attach a single bpmn drawing and maybe an architectural sketch to the process. Add time management, deadlines and maybe a tie in to the web services of an ESDH system and it might even work for task management in case working.
Everything is build for theoretical approaches. Like we do SCRUM, but really, we’re doing scrumish things. We have an odd schedule, we work on multiple projects at once, depending on what resources are available and what has higher priority, sometimes something breaks and then we’re all doing operations rather than development, sometimes the mayor has a direct request and so on. I think we’ve tried all the tools from atlassian to trello and nothing fits, it’s all too textbook for a messy place like ours and often I think we should go back to postits and a fucking excel schedule but I really don’t want to ever print an excel sheet ever again.
Interestingly I do a lot of networking with other managers in the public set for, and everyone had this problem, not just in digitization. There isn’t a single efficient tool for managing your workforce in the public sector.
There are excellent tools, don’t get me wrong, but we can’t have our workers spend hours on them because we can’t sell those hours to anyone.
We've gotten a lot of feedback that often times the strict structures of scrum and kanban are overly burdensome, yet teams still want and need some basic guardrails (as well as the ability to modify their processes on the fly). Our Product team is still testing and iterating on this new project type quite a bit, and if you're up for it, we'd love to give you an early demo and get your honest thoughts and feedback.
If you're interested, please shoot me an email and we'll find some time for a demo: jake@atlassian.com
Jake
Jira PMM @ Atlassian
Originally we used JIRA, but it was complete overkill. We've settled on making adhoc GitHub Project kanbans and very creative use of the labels, and so far it's been ok, but not perfect.
The other thing we have to do is time tracking for specific tasks depending on the project/client, which has us going over to ConnectWise (the primary part of our company is an MSP), which is just terrible.
The final problem is tying all of the documentation together. The MSP side of the company uses ITGlue, but it's not enough for everything we do as developers. Confluence was actually nice for that, but since we've left Atlassian we're just tracking stuff using a doc folder attached to the project source itself.
I've never used cushion, but I like clubhouse a lot.
As you can see the bulk of the problems in the health care industry is understanding how to navigate the huge mess of the US healthcare system. A longer term solution is for us to build a single payer system and incentivize patient care over patient procedures but I doubt that will happen.
I do think Zocdoc is a great idea because anything that makes scheduling an appointment easies is a win. I'm just saying that patient reviews give people a false sense of security that their doctor is good. In reality, almost no one knows if someone is good or not unless you have directly worked with that doctor and have a lot of stats on their outcomes (which only anyways works on specialities that have a lot of procedures). Assessing doctor quality is a very hard problem.
Patient reviews are good for assessing if your personalities will mesh but beyond that they don't go far.
I am not really too familiar with Oscar; what are they doing?
We actually explored this for a startup idea. The problem is that it is difficult to find a group of people who can do the evaluations in a truthful and holistic way:
* Hospitals will never want to give out outcome data because outcome data will be used against them for ratings by people who don't understand it (for example, some community hospital in Montana may be rated higher than Mass General because of case complexity issues). Or worse, it will be used by people who DO understand it :D
* We explored having doctors rate other doctors in a variety of ways (which I think would reflect the "truest" measure of quality). Residents and fellows could rate attendings, but they might not know how attendings in their hospital compare to attendings in most other hospitals. Additionally, attendings or hospitals might apply pressure to these groups to provide good ratings. Specialists could rate other specialists in their field, but then you might see collusion, false negative reviews, or retaliation. How you would avoid these problems is not immediately clear to me.
* As you point out, patients are really only able to evaluate bedside manner and not quality of care.
One way we thought about it is having a rating system which would have public profiles for physicians and anonynmous reviews from other physicians and members of the care team. Ratings coming from other physicians in the specialty and providers at their institutions would be weighted more than ratings from other physicians. The highest rated physicians would also have more weight within their specialty than an average rated physician. You could bootstrap the system by asking specialists to provide the names of the top X people in their field. These people would automatically be rated highly.
Patients could log in and provide comments about patient experience; hospitals could log in and provide outcome data if they wanted.
I am not entirely sure how you would really monetize it. It's the equivalent of the dating app problem; the better you are at matching, the less money you make as users exit the platform. I do agree that it would be great if osmeone could solve this problem though.
> Helping doctors and patients estimate costs - Neither doctors or patients understand costs. it is very routine for a doctor to suggest getting a lab test from X place because they have experience working with the center. They have no idea that for your specific insurance plan this will cost you 2x another place and so you with an unhappy patient who blames their doctor for ripping them off. There should be tools to patients and doctors estimate costs.
We had a startup that came at this in an indirect way. We were trying to make it easier for labs and other providers to perform eligibility checks and facilitate prior auths in real time. Our proposed solution would have involved running the check during ordering, and then having the phlebotomist or lab tech at the hospital doing the sample collection contact the doctor and inform them that something would or would not be covered by insurance. The provider could then talk to the patient about out of pocket costs etc. I think this actually could have worked; our team imploded before we could prototype it however :(
While I like the other ideas, glhf trying to get people to practice lifestyle management or be adherent to a treatment regimen XD
However, I have looked into incentivized, distributed platforms such as wings.ai or golem.network and believe that this could be applied to healthcare. If patients were incentivized on going to the doctor for a copy of the CPT/DGX codes on their bill and survey information around PQRS you would have a lot of useful data. This data could be used not only to provide a rating system of doctors, but also price comparison on CPT/RVU across the United States and the ability to provide PQRS data for doctors. Its an all in one, decentralized, healthcare platform for patients, doctors, and insurance providers.
As an example, look upon the bloodied corpse of the failed startup Remedy, which was actually doing something really good -- helping users find billing errors and getting money back for them on bad charges. But incredible amounts of pushback stymied them:
https://www.fastcompany.com/40483774/remedy-wanted-to-cut-pe...
Could a PMR system help with some of these issues? Giving patients access to their records making it easier to aggregate data abiding with HIPAA?
I recently saw an episode of Dirty Money on Valeant. Is Valeant a good representation for most pharma companies and also a reflection of most for profit medical businesses whose primary obligation is to increase shareholder value?
What would be the best way to start aggregating and comparing prices in the USA? If it isn't possible, why do prices vary so much?
Patient medical records won't help with pricing issues, but data transparency would help with automated solutions to manage care.
There is no easy way to aggregate and compare prices. If you could get billions of health care claims then you could build algorithms to estimate pricing or perhaps you could convince the large health care companies to share their pricing with you (highly unlikely). Prices very widely because there each institution negotiates their own set of prices with a specific set of providers. Basically imagine your health care plan is basically a set of discount codes for a set of doctors and providers. Except those discount codes are never shared with you and vary for each doctor and each provider and can change at any time.
I think this need, a real need that is, can potentially make millions for the enterprising type, here.
Why don't I try the same, you might ask. But, while I have some ideas, I am may not be able to raise the resources needed, at the moment.
For those that want to follow up on this, you could use current satellite imagery to make population estimates, but it wouldn't be really accurate. You could wait for companies like Planet Labs[1] to get daily photos of the Earth (IIRC 4m resolution), and increase accuracy. You could also hire pilots to strap a camera and survey the interested areas. There are also, IMO, unethical spins (especially in oppressive regimes) that you could do with that data (ie: civilian surveillance). Or even cheaper, use drones (because you don't need daily or real time surveillance). That could also have other spinoffs. Or you could literally just send people to go count. Don't know which would be the cheapest.
[1] https://www.planet.com
Some of the best stats you get for telecom coverage will come from the annual reports of commercial telecom companies. Eg in Kenya you can look at safaricom. In somalia I guess you can look at Telecom Somalia and some of the others (keeping in mind their market share).
GSMA do research every few years on the number of SIMs on average per market. This is how they get to a figure about subscribers (ie people) rather than connections (number of SIMs). The latter is what is usually reported by commercial companies.
You can quite easily do some online digging for a given market to get fairly current numbers (ie last year). For doing anything with sizing markets etc I'll always start here. Not with World Bank or similar data which is really outdated and not indicative of current trends in technology at all (which is important for me).
The international development sector is unfortunately so slow to realise what's going on technologywise - i'd wager Facebook will know a huge amount more about populations in a lot of these African countries than their governments or the international development sector do. Perhaps if they were a bit more genuinely philanthropy inclined they could do some amazing socially impactful stuff right now. But I don't think any of us will count our chickens on that one.
A mistake can mean - misconfiguring your target (wasting money on ads that won't give you an ROI), misconfiguring your 3rd party tracking (letting data like conversions go unaccounted for, or not having your auditing tags setup, meaning you show ads to fraudulent users that you otherwise wouldn't have to pay for), etc.
Alternatively, one place to manage creatives and language as well as targeting for each campaign might be helpful as we use google sheets, excel and Trello for this now.
The real issue is how custom these solutions need to be and which weird rules and workflows some need to follow because of internal rules, iso norms and whatever else.
I don't think a one fits kinda all solution would ever really work. However i highly believe that companies need to allocate enough time to build tools that work _for them_
Our process and system is super efficient recording video + sending notifications to team members to edit, add captions, strip audio for your podcast, set up your podcast, set up your alexa flash briefing, etc.
But it takes hours to do all of this if youre on your own and thats ONLY if you know how to do it all. Content is the black box most have no idea how to do. If you don't pay our agency to do it for you you are kind of out of luck.
We sell a book on our process now and sell about 50 copies a week. These people are validated and want to learn how to do it, and are willing to pay to learn.
It only makes sense to build the platform that automates this process for these people and offer it to them. They've already paid to learn. Might as well offer the platform to do it.
Edit: I'm in the computing industry. Have you heard of it?
its a $600B industry that is in decline because its traditional R&D engine is sputtering out, and big pharma has been amazingly acquisitive the last five years to replace off-patent blockbuster drugs (more IPOs and big M&A than software the last 5 years despite getting 1/5 of venture funding)
tons of really interesting new tech for startups to explore: synthetic biology, cell and gene therapy, bioelectronic medicine, many many others
Our marketing department sends 10+ campaigns/week, each goes through multiple changes and compliance approval. It’s very time consuming, especially when someone needs to touch the code.
> especially when someone needs to touch the code
Do you not generally touch the code? Is some kind of WYSIWYG editor used to create an email or is the code human written?
So, html email template (no unlined css) > merge word content in dreamweaver > upload to custom web app > app uses premailer to in-line the specified css > make changes to the content via ckeditor (which messes up the html in many cases) > export in pardot-ready format.
The app versions the content changes and allows comments as well easy email tests. We do 20+ revisions per campaign so that’s a necessity.
This was built 5 years ago and still works but getting inadequate for the complexity of responsive emails.
I’ve looked at this: https://beefree.io/ but it does not allow the creation of custom content modules. So looking into building a new version of our own, again.
It’s a pain but nothing compared to debugging email markup.
When installing these at scale in existing buildings you have to be able to send out local workforce to properly install and activate thousands of sensors, as well as maintain them afterwards, without prior training. It’s one of those things that sounds easy on the surface but is riddled with complexity, like how to register which sensor is installed where in a foolproof way, or how to easily locate faulty sensors for replacement.
There’s plenty of competition in people selling the sensors, providing connectivity, or doing data analysis, but I’m not aware of any solutions for installing the damn things which aren’t tied to a vendor.
Maybe it's just a feature, not a full product, but it makes any "learn to code" MOOC unusable.
Jest and Mocha both have a watch feature that will test a file each time it is saved for "continuous testing", much like using Gulp watch or any type of dev live server environment.
Also Jest features Snapshot testing which will take a picture of a UI and test all changes made to the UI, as well as alerting you in tests if the UI has changed. I could see this being used in bootcamps as well.
you install a cli program, write the code, type something like: learn test .... tests are ran and results show. if all tests pass then dashboard is updated showing tests are passing.
Secondly, with VPN, I first have to connect to the open network in order to activate the VPN. I also need to do it for each device I want to connect (phone, computer, tablet).
If you are logging into services that can identify you, checking your personal email over a TOR connection, or doing work over a TOR connection, you're putting yourself and your company at risk.