Launch YC S21: Meet the Batch, Thread #9
This is most likely the last batch thread of S21 startups. The previous thread was https://news.ycombinator.com/item?id=28233916. The original description is at https://news.ycombinator.com/item?id=27877280.
There are 4 startups in this thread. The initial order is random:
Snowboard Software (YC S21) - Fast and automated data catalog for Snowflake - https://news.ycombinator.com/item?id=28315047
Jovian (YC S21) - Online data science school for professionals - https://news.ycombinator.com/item?id=28315045
Chari (YC S21) - Ecommerce and fintech app for mom and pop shops in North Africa - https://news.ycombinator.com/item?id=28315043
Tuli Health (YC S21) - Turning UK pharmacies into diagnostic centres - https://news.ycombinator.com/item?id=28315046
51 comments
[ 4.3 ms ] story [ 127 ms ] thread80% of the FMCG (Fast Moving Consumer Goods) consumption in North Africa, and in developing countries in general, goes through small mom and pop stores (Vs 20% in the modern trade, such as Supermarkets). We are husband and wife and have two kids. We live together and go to the same mom and pop shop next to our place. Many times, while we were buying our groceries, we were left waiting because the truck of Coca Cola (or any other supplier) came by and the shop owner needed to deal with the supplier instead of his clients. When we ask the shop to serve us first, he replies that he can't because the supplier vans companies can't wait. Indeed the streets are narrow in North Africa, the vans usually create traffic jams and they have to sell and unload very quickly while the cars waiting behind are horning.
With Chari, they don't need to let their clients wait. When they run out of stock of a specific SKU, they add it to the app during their spare time between two clients. At the end of the day, they have the list of the SKUs that are missing in their shop. They then make the order to be delivered the day after.
We launched 18 months ago in Casablanca Morocco and have been growing at a 40% monthly rate on average. Last month, we opened our second warehouse in Morocco and our first one in Tunisia. Here's a promotional video if you're curious: https://drive.google.com/file/d/1HEZla3KvtWYM06PO94igppL95ea....
btw, congrats on the launch. I consulted for similar startups in india and everyone is trying to get into the "kirana" model
You note the idea of them ordering every day, are the goods moving that fast that it makes financial sense for daily orders? Normally these shops are ordering weekly or biweekly.
Do you work with the Principals at all (e.g. Coca Cola) or just directly with the Distributors?
Did the merchants already have smart phones and digital literacy to be able to easily download and leverage your app or was it a process of teaching them how to use it?
We are using a transport management system to optimize our daily delivery routing. Shops are really close to each others , so it makes financial sense to daily deliver the same shops. There is no minimum order , however some elementary SKUs can’t be bought only by themselves .
We get our supplies from both the Principals and some exclusive distributors.
We start first by sending salesrep on the ground to train the users on how to use the app. Usually the first orders are made through our customer care team. It can take up to 3 months for a client to become autonomous and use the app by himself.
Four years ago, we quit our jobs and spent several months learning data science and deep learning. We completed several online courses, read some books, and participated in some Kaggle competitions. We were finally able to make a career transition to data science, but looking back at the journey, we felt “It shouldn’t be this hard!”
While there are various platforms offering data science education, most learners don't feel confident while approaching real-world projects even after completing multiple courses and earning certifications. They find it hard to clear interviews to land their first job, and often require several months of training on the job.
Jovian is the platform we wished we had access to when we started learning. There are three things we focus on: (1) we teach practical skills that students use to build real-world projects, (2) we offer 24x7 support & detailed feedback on every project, (3) students learn together with a community of like-minded learners in a university-like environment. We’ve taken inspiration from courses like fast.ai and mlcourse.ai.
Our 6-month program consists of 7 courses, 14 coding assignments and 4 portfolio projects. Every week, students watch 5-6 hrs video lectures and work on assignments for 15-20 hrs. Students can ask questions and get help 24x7 from our TAs over Slack and Zoom. Every 5 weeks, they complete a unique guided real-world project and publish a blog post. We offer 1:1 resume review, interview practice, and job search support for 12 months. We make money by charging a flat fee for the course, which students can pay up-front or monthly.
One thing we discovered and found surprising is that most of the math used in data science can be explained through code. Many people find math intimidating not because the concepts themselves are tough, but because they aren't used to mathematical symbols and equations.
We look forward to your comments!
I know that YC has also funded Lambda School, which offers a data science course, but their payment model is based primarily on ISAs. Did you consider ISA payment plans, and if so, why did you go with the flat fee? If not, do you think you would want to move to an ISA payment system as you scale?
Congratulations on the launch and best of luck!
Our program is designed primarily for working professionals with some background in programming, analytics or statistics. We've found that most of them are comfortable with paying program fees upfront or monthly.
However, we do also offer an ISA option in cases where students are unable to pay upfront.
Have you thought about selling courses into large organizations? e.g. "SQL + Statistics for marketers", I would love to see something like this.
I always thought of Jovian as a lean business but this surprised me. Also are you planning to build the Notebook-like platform (that it was before) more or the future is only data science bootcamp?
Our goal is to offer practical & affordable data science education to millions of people around the world, and joining Y Combinator is going to help us execute on our vision faster.
We launched the bootcamp because we saw a need for a more hands-on and intensive program to help our users make a career transition to data science. However, we will continue to offer more free courses and improve the notebook-based learning platform.
In all seriousness though, VC funding is mainly a function of market size and scalability of the business model.
There's currently a huge shortage of data science talent across the world, and a sustainable business funded by venture capital can help fill this gap quickly.
On your website homepage, you claim 160,000+ projects. What's the definition of a project in this context?
Students who are part of our 6-month program build guided portfolio projects that require at least 30-50 hours of work and must meet our evaluation criteria.
I see your point though, perhaps we should use different terminology for the two types of projects.
I have a MD PhD from University College London, and Jialu has been a software engineer for the past 10 years. We’ve lived in London for a decade and while we appreciate the NHS, we also wished for a world where we could walk down the street to get healthcare. This is what we're building.
Instead of travelling a long distance and waiting a long time to receive diagnostic tests, customers can now receive diagnostic tests from their local pharmacies, just around the corner from where they live.
There are 11,000 independent pharmacies in the UK. Our software makes it easy for them to sell and fulfill Covid-19 tests. These same pharmacies have just started selling allergy tests as well.
We offer easy-to-use software for the pharmacies to register and manage patients for diagnostic tests. We are currently used by over 150 pharmacies. The link to the dashboard is https://portal.tuli.health. A user can sign up as an end customer or a pharmacy.
Covid has led to changes in our user behaviour. Many services have now become hyper-localised. Covid tests led to the pharmacies delivering the first diagnostic tests. In the future, other diagnostic tests will also be available: STI, , biochemistry profile, gluten intolerance... Hopefully things will get easier and more accessible. We look forward to your questions and feedback!
I'm curious as to how you will integrate into the existing UK healthcare system. Will you have access to my NHS records? Will the results of these tests appear on my NHS records? Will these tests be covered by the NHS?
(I'm not in the healthcare profession so please forgive me if these questions are obvious to those that work in the UK healthcare!)
The marketing copy on the website seems to suggest this is for private healthcare ("All your private healthcare needs in one app")
I will be interested to see how this goes, because the sentiment outlined in the OP can be quite controversial in the UK healthcare market ("while we appreciate the NHS, we also wished for a world where...") - for international context, the NHS employs 1 in 17 people in the UK, and there is an almost irrational love for it. The UK's initial covid lockdown message even featured it as an imperative - "Stay at home. Protect the NHS. Save lives". A push towards private healthcare (at point of use) may well be a challenging sell, though perhaps a primarily-urban market might accept it more.
I'd have expected this to go down pretty poorly because while high earning busy people tend to be more open to spending to fix problems, the idea of paying for healthcare here is nuts – you either use NHS services, or your employer gives you private healthcare.
I wonder if, as London becomes increasingly diverse with more workers from overseas, the acceptance will increase as there are more people who don't have that cultural blocker to paying directly for healthcare (rather than via taxes).
Some seemed to be trying to deliver NHS services (video/remote consultations available by "gig-working" GPs in evenings etc.), but others seemed to be increasingly trying to provide testing or other preventative/diagnostic services (blood tests, 7-day health check-ups booked same-day etc.)
I guess to some extent there likely will (among high earning, time-strapped immigrant workers) be an increased appetite to pay for healthcare directly, but for those who are aware of their immigration NHS access charge, they might well come to adopt the "I already pay for this" view. There's also a scalability question beyond London, where typical incomes mean the only people using any kind of private healthcare will be those with company-paid schemes.
There are some interesting ethical quandaries around what the NHS could do to improve its service (which could potentially hurt private sector rivals) - from some people's perspectives, the NHS could allow people paying privately to get things faster (without compromising on clinical need based priority) if it meant the service was better funded through revenues from those private patients going towards cutting wait times more generally. The British pastime of queueing could negatively impact on perceptions of anything like this though (even though you can already pay a little extra for an expedited next-day passport renewal etc.)
I know some hospital trusts have setups with private wards where NHS staff deliver private care (without bypassing priority by clinical need), and revenues from this are reinvested into regular services.
In past jobs, I (Rick) rolled out Snowflake at a manufacturer with 11000 employees and $2B in revenue, and Théo built the data team and infrastructure at a B2B company that raised $200M. My experience introducing Snowflake was that after 2 years and more than 1000 tables and many more columns, it was hard to find data and tedious to understand it, and users would not trust the data without a lot of upfront due diligence.
We index all data and visualization assets that are connected, making it easy to find them. We profile data continuously to show what's inside, how it is changing, and make it easy to document and tag information. What helps us surface a lot of context is that we parse Snowflake's query history. This lets us provide automated data lineage and indicate data freshness and usage.
Most of our customers that see their actual data flows are surprised by the complexity they have built. Some see for the first time how their data flows work and how many redundant data assets they have created from one base table.
For now, we focus on the Snowflake ecosystem and don't support the long-tail of other database and analytics solutions. This lets us "overfit" to the needs and problems of companies that sit on this stack, e.g.: Fivetran > DBT > Snowflake > Tableau. Under the hood we leverage the flexibility of ArangoDB to offer a fast application that offers great search speed and also graph workloads e.g. for data lineage.
Our initial market is 5k Snowflake customers, but the global data warehousing market consists of 250k companies. We charge per indexed data asset. The number of data assets in companies is doubling on average every 2 years, so this market is taking off.
We offer both cloud and self-hosted. Self-hosted means a Docker Compose configuration that starts 3 Docker containers. The only data that is transmitted to our system is a license check in this case.
Your questions and comments are welcome. Thank you!
I ask because I built a SaaS product that helped customers understand their data inside a legacy accounting product. Similarly it looked in the meta data on the back end of that accounting product and built visualizations. The problem we had was that customers didn't know how to or didn't want to fix the problems that our tool found. We had to hire accountants to go into the accounting tool and fix their problems. So it ended up being more of a tool for our accountants and less of a stand alone SaaS tool that customers actually used.
Further down the road I can also see that consultants will find it useful to help their clients clean up their data pipelines and reduce complexity.
Column statistics: Sample data randomly and calculate estimator from that.
Data Lineage/usage: From the queries (and 'similar' stuff like materialized views. It seems like that currenltly have only lineage on table level, i.e. which table depends on which tables. Do you plan to implement data lineage on column level, i.e. which column depends on which column? This sounds challenging and very interesting.
I tried to watch the demo video on your website, but it was too small and maximizing the video is disabled (Firefox). So I had to click through to youtube. I'm not sure if this is intended.
Column Level Usage is something we are looking into. We also have some ideas. It seems like a solvable problem but as you said challenging. :)
Does snowflake have any plans of building a similar cataloging infrastructure?
To answer your question: there is multiple parts to it, I am certain they will tackle some parts like search in the future. I don't see them solving lineage or data monitoring anytime soon. Also cross-system overview (e.g. with Tableau) won't be coming.
To give an opinionated position regarding Amundsen:
Amundsen is a great open source project with many cool ideas. However, we found that the experience on Snowflake was quite shallow and much more insights and context on data assets are possible. E.g. data profiles, lineage or usage information are not readily available in Amundsen. You can customize and extend Amundsen to do more, but most data teams should build and innovate in other areas closer to the business. ;) Most companies use Snowflake over the Apache Spark open source stack exactly because of that.
If you are interested to learn more, please reach out: rick at snowboard.software
Edit: here’s their YC company profile https://www.ycombinator.com/companies/mentum
Is the best path to forming a startup these days to take an already existing product and market it to a very specific niche while taking investment capital that rivals the revenue of pre-existing software?
I don’t get it. Really.
I like using products that were created keeping me in mind as the primary customer. Wouldn't you too?
I suppose it's mostly an artifact of YC's growing so more internationally in recent years. There are others (e.g. https://news.ycombinator.com/item?id=28257352 and https://news.ycombinator.com/item?id=28087309) but these are a small minority.