In September 2011, Allen and I started building a PaaS. With two people. And not much money. In just over a year, our tiny two man team built an autoscaling PaaS for testing, that is used by thousands of developers to test code every day.
Our little PaaS is pretty good: CircleCI is by far the fastest way to run your tests (faster than your own servers, EC2, or other CI services). It offers facilities that don’t exist anywhere else, like the ability to SSH into our machines to debug your tests. And it automatically sets up new projects by inferring test settings, and splits your tests across multiple machines to get your code to customers in a fraction of the time.
And at the same time we built a business: from receiving our first payment in April (thanks Zencoder!) we now have thousands of developers paying to use CircleCI. We help big teams like Kickstarter, Stripe, Red Bull, and TapJoy scale their dev tools. And we particularly love how great products we use ourselves are tested using CircleCI: Intercom, Airbrake, MongoHQ, GrubWithUs and Stripe.
Can we do better?
But we started seeing the limits of what we could build with just two people. The fast, in-depth support we provide takes a lot of time. Scaling our service, adding features to make you productive, writing detailed docs, that all takes time. We listened to your feedback and the kind of things you want from us, how to make your development lives better, and we saw what it would take to build them.
We make enough money to bootstrap, but in order to really provide the kind of product we wanted, and the sort of things you asked for, we decided to raise money.
As soon as the first angel checks started coming in, we got straight to hiring. We posted a few months ago about David, our first engineer. Since then we’ve hired Daniel, who shipped the personalized dashboard and the branch view. We hired Jenneviere to keep the bills paid and lights on as we build out the platform. And we hired an actual designer to create the experience we had always intended but didn’t have the skills to put together! Eric started last week and redid the homepage, and will soon be turning his attention to the core of the app.
[We didn’t invite Eric to the Zoo]
We’ll be hiring at least four more devs in the coming months, and though our pipeline is full, please drop us a line if you’re an awesome engineer who’d like to play with PaaSs, Clojure, and beautiful, usable front-end apps. The CircleCI platform is a joy to code, we have a great flat culture (we’re trying to emulate Valve and Stripe, in our own ways), and we think you’ll really like it here.
The actual funding news
We raised $1.5m. We’ve got a great product, a great reputation and great customers, so we got great terms. And we got great (and very technical) investors: Heroku founders James, Orion and Adam; Heroku investors Baseline and Harrison Metal; Valley favourites SV Angel and 500 Startups; and Eric Ries whose company IMVU invented half the technologies we’re making, and whose Lean Startup taught us a lot about how to build software and businesses with small incremental changes.
We also got some wonderful angels on board. Hiten Shah runs Kissmetrics; Techstars’ Jason Seats made Slicehost; Jonathon Siegel built ELC, RightScale, and now runs Airbrake and RedisToGo. Data Collective’s Matt Ocko invested in most B2D companies worth talking about; Ronnie Cheng is an former Facebooker who now uses CircleCI every day; Issac Roth runs PaaS at Red Hat; Dave Concannon is building his own B2D startup HostedGraphite; Lars Kamp runs Accenture’s Corp Dev dept, and Colm Doyle has started teaching me all about enterprise sales.
The major reason we raised is simply to make the product better. We have lots of customers who love CircleCI, but there’s more we can do. We talk to our customers all the time, and so we know your problems, and that’s what we’re here to solve. And the major blocking factor was being just two developers.
We’re now four full-time devs (I’m ashamed to say I don’t get to code as much as I’d like to), and we’ll be eight within a few months. We came as far in the last few months since David and Daniel got here, as we did in an entire year before that, and there’s more coming.
So what are we working on to improve your lives and make you happier? As always, your productivity is our number one goal, and we have a couple of lovely features coming out that will allow you to ship better code faster.
Firstly, parallelism is going up. We have lots of customers who have long test suites, and even with the raw speed of our platform, some folks always want to go faster. We’ve had lots of customers using our 4x parallelization, and we’re starting to increase that to 16x, and see how far we can take it. Imagine the change to your feedback loop when your hour long test suite runs in just minutes.
We’ll also have a heavy focus on Continuous Delivery. Some of the advances in CD from Facebook, Etsy and IMVU would make us all a lot more productive. Things like atomic deploys, the ability to rollback deployment in response to metrics (like losing money, or an increased error rate), or management over staged rollouts and feature flags. All of this is on our roadmap, all with the aim of letting you ship better code to your customers.
And of course, you’ll be getting more of the same. Our machines will get faster, support more software versions, be resilient to upstream failures. We’ll expose more test data to you, expand the API, detect more problems in your tests, and give you more ways to debug failing code.
Thanks again for using Circle, and keep the feedback coming. We’ll keep making tools to make your job easier, to ship to your customers faster, to spend more time writing software, and less time testing it.
Paul, Allen, Daniel, David, Eric and Jennviere