Fresh Insights: YC W15 Developer-Oriented Company Cheat Sheet
Disclaimer: I’m a founder of a YC startup (FlightCar).
At FreshVC, one of our biggest areas of interest is technology that fuels the technology revolution — check out our investments Aptible and Vessel. I’ve outlined some basic information and thoughts on developer-oriented companies in the upcoming YC batch.
As my partner Brian mentioned in his post, “YC W15 Marketplace / On-Demand Cheat Sheet”, I hope this will help investors have more meaningful conversations with founders and encourage data/thesis/belief driven, instead of hype-driven, investments.
Readme.io enables developers to quickly build beautiful, interactive documentation for their APIs. Their current offering is quite robust and provides a better experience than most can build in-house for their APIs.
Readme currently offers a $14/mo and a $59/mo plan, both targeting early startups. They also have a custom plan for enterprises and larger startups with no visible pricing. I think the real money to be made is in the services beyond just their current offering — once they’re in with an API, they can score big by providing auxiliary services, eventually they could perhaps even create APIs for companies. It will be interesting to see the direction they choose to go, beyond just documentation.
Readme is well-respected within the developer community for having a beautiful, well-designed product and website.
While Readme has built a robust solution for documentation of APIs, that alone isn’t a massive business — Readme has positioned themselves well to foray into overall API management (access, sandbox, reliability, etc) and appears to have started that transition with the advent of API Key management. This brings the potential of the company up to the levels of Mashery and Apigee who will probably be their long term competitors.
Treeline enables anyone to build a backend for their website using a visual interface. They can also host the website using Treeline’s integrated hosting platform (currently very beta). Notably, the company also created the popular Node.js framework Sails.js.
Unknown, likely can charge a multiple on AWS for monthly hosting in the future — I feel like the heroku-like freemium approach will work well here (first dyno free). A Parse-like approach might too (requests/s).
Treeline is working on a huge problem — making programming more accessible to the public. The potential is large, the product looks promising but is yet to be proven.
It will be interesting to see how quickly the product is adapted by users in the general public and how they can promote the ability to program visually beyond their beachhead market of designers/front end engineers. Their Product Hunt is the 16th most upvoted of all time, which is great given that Product Hunt users will likely be their early adopters. Beta users’ twitter comments about the product have been overwhelmingly positive as well.
Gitlab is a competitor to Github enterprise, offering not only code version control but code reviews, an issue tracker, a wiki, and a continuous integration/deployment system. Gitlab, as it is, is already a very full-featured and substantiative competitor to Github, providing all of the basics to own the management and deployment of a codebase in a company (traditionally handled by stitching a suite of software systems together, e.g. Jira for bug tracking, SVN for version control, Teamcity for deployment/CI, etc).
Basic: $19.90/user/yr, in multiples of 20 users
Standard: $49/user/yr, in multiples of 100 users
Plus: $249/user/yr, in multiples of 100 users (this seems to more of a gimmick tier, promising “prioritizing features you want” as a benefit)
The standard offering is most comparable to Github Enterprise which costs about $250/user/yr, making Gitlab a steal in comparison.
What’s really interesting about Gitlab is that they’re aggressively competing with Github by providing a lot of paid features on Github and other services (like Circle) for free to the community: private repositories, free continuous integration, and unlimited disk space.
According to their CEO, from a Quora Post, over 100k organizations use Gitlab — if that’s true, that’s a massive, very valuable user base. On their website, users include behemoths like Nasa, IBM, Expedia, Alibaba, At&t, SpaceX, and Qualcomm.
As a developer, Gitlab’s offering is very clear, and it seems to be essentially a cheaper, more integrated version of enterprise Github. I’d be interested in seeing what their free tier costs them, how much enterprise customers are paying them today, and the extent to which these enterprises use them. I’m also interested in hearing someone’s experience using both Github and Gitlab — I can tell you as someone whose used alternatives to the two, these two are the only two I’d consider within my own organization.
Outbound is a new A/B testing and notification service that enables developers/PMs to test and send push notifications, emails, and SMS based on triggers by users (e.g. signed up, reached check out page). Mixpanel does a little bit of this in a more manual way — you can send push notifications to your cohorts, though Outbound is definitely a lot more robust as it’s focused on the problem (making it a good Mixpanel, Salesforce, Marketo, etc acquisition target in the short term).
2,000 active users: $149/mo.
10,000 active users: $399/mo.
30,000 active users: $899/mo.
Looking at Outbound from the lens of a developer, if I need to do trigger based messaging & notifications, and these messages are core to my business (maybe a drip oriented enterprise sales campaign or a consumer online retail company), Outbound makes it substantially easier and cheaper to build and update these components than it would be for an engineering team to do so, in coordination with the PM and business unit of the company.
They’re also solving a part of the last mile of analytics, analyzing the potential of the bait that brought a user to the site, not just the flow of the user through the site. I’d be interested in seeing what common use cases of Outbound are and what their sales process is like/who they’ve sold.
Seed is a new type of business bank which enables businesses to move very quickly — allowing for fast transfers and full API access to all actions, groundbreaking in an industry with frustratingly slow processes for businesses.
Seed is in beta still, so no revenue model released — but the sky’s the limit when you store all of the money.
Seed is tackling a major problem in a legacy industry. Using Seed or Standard Treasury (also YC from older batch, same exact model) is almost a no-brainer for developers looking to programmatically manage accounts and transfers within a system cost and speed efficiently. Generally the problem with programmatic banking isn’t the API (though these aren’t great) as much as it is the infrastructure that is a bank. Seed especially seems valuable to marketplace oriented businesses where several of your users will be paid out in funds via ACH/Wire.
I’d be interested in hearing about how they’re different from Standard Treasury and more about the different use cases and whether they’re focused at all on users who wouldn’t benefit as much from their APIs but would from the quick-turnarounds they promise. It’ll also be good to figure out what their exact revenue model is and what they plan on doing with the cash they’ll be sitting on.
SigOpt is an optimization as a service platform, offering optimization at a level-deeper, from A/B testing to Machine Learning (recommendation systems, categorization, etc) to physical ingredients in a product formula (saving time synthesizing noisy formulas).
$0/month for 1 experiment,
$99/month for 3 experiments,
SigOpt’s offering is really interesting as ML is very inaccessible to most developers today, let alone consumer product companies, drug synthesis labs, etc where it isn’t an expertise.
The concept of being able to pass SigOpt a dataset and in return get categorization, recommendation, optimal flow by cohorts, etc is an incredibly powerful one, and the system they’ve built isn’t only powerful because of the science it replaces but also because of the operational complexity of analyzing datasets like these, abstracting the scale of servers, the computational optimizations, etc to turnaround results to a much less sophisticated user, quickly.
Yhat creates two products, Science Box and Science Ops, both of which aim to make data science workflows as easy as developer workflows have become in the last few years.
Science Box is a development environment for data scientists, offering a cloud IDE and a collaborative platform for teams as well as a layer over AWS, enabling data scientists to utilize horizontal scaling in the cloud in an easy, abstracted manner to run their programs.
Science Ops is more targeted toward enterprise customers abstracting the scaling aspects of Science Box within an enterprise’s personal cloud or machines. It also seems that Science Ops is the tool used for deeper integration within an application, having a richer API — though I’d want clarification from the founders on this point.
Right now, they’ve only listed an Enterprise model. Probably can make a multiple on AWS or a per server fee at the private deployment level.
Yhat is tackling a very necessary problem — programming complexities and infrastructure get in the way of being a data scientist. Yhat has the potential to make data science a lot more accessible, growing the market as more people understand it, by creating simple tooling around it, similar to what companies like Heroku have done for web development.
I’d really be interested in learning more about product usage from Yhat and conversing with existing customers about previous workflows vs Yhat’s current one. I’m also interested in learning more about their pricing model and who they see their competitors as.
Pachyderm enables developers to write large-scale map-reduce tasks (think fundamental blocks of big data processing) using Docker on a CoreOS cluster rather than Hadoop or your own custom solution.
Not yet decided, probably can charge a multiple on AWS for distributed compute jobs.
Pachyderm is making Map-Reduce a lot more accessible to today’s developers. Previously, in order to write a Map-Reduce job, you’d generally use Hadoop which requires developers to program with a Java interface, unfamiliar to many, though there do exist libraries in other languages to access Hadoop.
Pachyderm enables developers to use simple, dockerized http servers instead, meaning they can use the programming language and corresponding libraries of their choice, making Map-Reduce not just more powerful but easier and modern. Being able to use any programming library is a huge bonus, as with Hadoop, existing libraries are generally clunky, in JAVA, and far fewer than the suite of libraries available throughout the internet. Because Pachyderm uses docker with an HTTP interface, it is programming language and library agnostic.
It will be interesting to find out what their end revenue model is. Pachyderm isn’t yet fully production ready but what they’re building is very hard to build and needed — in the banking industry for example, where data sets are very large and money isn’t an issue, large engineering teams often build out their own custom solutions because of the limitations of Hadoop (language support, etc). Pachyderm eliminates that need.
Omniref is creating a reference for every open source library that is hosted on Github using user-created annotations — think RapGenius for code.
Unclear as of yet, Stack Exchange does job posting and engineer database sales to generate revenue.
One of the most interesting parts of Omniref is its social application in the programming world. Since Omniref connects with a user’s Github, it’s able to show her the most relevant annotations/comments to her existing work, I’d guess better at engagement than Stack Overflow for that sake.
Omniref is also probably using the same SEO techniques as Stack Overflow and Rap Genius with its rich user-generated content; I’d guess by the way Omniref’s site works that it’s an SEO gold mine, since it hosts every open source line of code as well as relevant annotations next to them.
Since Omniref is inherently a social platform, I’d like to learn more about how the founders think about engagement, churn, growth, and future features. I’m also interested in learning about their plans to monetize the product and what their thoughts on enterprise offerings are.
Akido is a platform that deeply integrates with hospitals’ existing medical systems in order to provide developers with a generalized API to easily make Health IT apps. Its competitor Apervita does the same thing, though it also forces developers to be part of its existing app-ecosystem (basically trying to create Google Play for Health IT). The company is really similar to Clever, which solves the problem for the education vertical.
Neither company shares their pricing structure publicly. My guess is Akido charges for the API and Apervita does revenue share with the app developers.
I think Akido is in a better position than Apervita here as developers hate building and being restricted by an ecosystem (just ask any developer how they feel about the App Store publishing process). Akido takes a better, more developer-friendly approach by just providing an API and being agnostic from there.
Akido has already signed on 200 hospitals, a non-trivial amount. I’d be interested in hearing about what apps currently use Akido and what the hospitals expectations from the integrations are. I’m also curious about how long integrations take and their long term revenue model, as well as their sales process.
Razorpay is the Stripe for India, enabling developers with India based companies to quickly and easily integrate payments in to their app, without dealing with hassles revolving around the bank.
2.5% per transaction for startups, 2% per transaction for businesses
Razorpay is clearly in a massive and fast growing market, providing core infrastructure to starting an Indian e-commerce company, fulfilling the need that Stripe/Braintree fulfilled a few years ago in the USA. Interestingly enough, Stripe does not yet support Indian companies nor does Braintree; Paypal is the most modern payment gateway that enables Indian companies to accept payments, but it owns to much of the payment process for developers’ tastes (redirecting you to their silly website to finish the payment, etc).
Razorpay is realistically the best payment option for Indian startups to use. I’d be interested in learning about: who is using them currently; how they see themselves competing with Stripe when Stripe enters the market; what prevents Stripe from entering the market; and whether there’s something fundamentally different about how Indian companies want to use a payments API vs American ones.
If you liked this post, please recommend and join our mailing list to be informed of more!