From Filepicker to Filestack: building a files API
Welcome to GET PUT POST, a newsletter all about APIs. Each edition features an interview with a startup about their API and ideas for developers to build on their platform. Want the latest interviews in your inbox? Subscribe now.
This edition, I spoke with Pat Matthews, CEO at Filestack, and their developer evangelist Dylan Simpson. Filestack began as Filepicker.io in Y Combinator’s summer 2012 batch and focused on making it easy to upload files from any cloud provider. Earlier this year, the company changed its name to reflect its expansion into a suite of file infrastructure APIs.
We dove into specifics on the business (their annual recurring revenue and growth), challenges of rebranding, and marketing experiments. Enjoy the interview!
Tell me more about your background and how you came to Filestack.
In 1999 I dropped out of Virginia Tech with two of my best friends to pursue our dot-com dreams. Three years and two pivots later we found ourselves building a SaaS company before SaaS was really a term.
In 2007, when Rackspace acquired Webmail.us, we were on track to do $8m in annual recurring revenue, growing 100% per year. I went all-in at Rackspace and spent six years there, most of which I was on the senior leadership team. Rackspace was pre-IPO when they acquired us and it was an amazing experience to be a part of a company going public and growing like we were. I wouldn’t trade it for anything.
Filestack predates me by several years. The original company, Filepicker, was incubated inside Y Combinator. We didn’t change the name until earlier this year. The founders had a vision for the idea that users are no longer storing files on their hard drives, so they created a developer service that made it really easy to integrate a file uploading service from any cloud service on the Internet. If you think about it, Dropbox, Facebook, etc. these are the new hard drives.
In 2013, Filepicker was acquired by Xenon Ventures, a small private equity firm in San Francisco, where it grew and developed for two years. Then in 2015, it changed hands again, and is now a Scaleworks company. Scaleworks is a startup venture equity firm based in San Antonio, where I live. I’m good friends with one of the founders of Scaleworks and that’s how I got involved.
The Scaleworks team really saw a vision for what Filepicker could grow into. Customers had pulled the company beyond file picking over the years but the brand and messaging had not evolved with the product. This is why we changed the name and repositioned the company in the market.
We really want to help developers manage files, images, videos — any kind of rich media on the web. We see ourselves as a powerful API for end-to-end file management. We’re trying to become a real player in what I see as an emerging space for developer services.
How big is the team?
We’re about 25 people now. We’ve got 15 people here in San Antonio, and this is where we’re growing sales, support, operations, and starting to do some product development. In Krakow, Poland, we’ve got a 10 person development team. They’re working on a lot of the back-end infrastructure and API.
What can developers do with the Filestack API?
Developers don’t want to recreate the wheel. We have a very polished API that lets your users upload files from anywhere.
The team’s original vision was to grab files from any cloud and store them to any cloud. That’s really powerful, but what we found out along the way is that people don’t want to just grab these files. Stripe is the poster child for handling billing infrastructure, and we want to be your file infrastructure.
Stripe was previously featured on GET PUT POST:
People don’t want to just upload and download images, they want to manipulate, crop, and filter them. There’s all kinds of tech that we’re developing and partnering with other companies on for image transformations. For example, there’s the whole Google Vision project that we think is pretty cool.
By the way, 90% of the files that we interact with are images, but we support files of all kinds. One of our big differentiators is we support document conversions, PowerPoint conversions, etc. We can really handle any kind of file.
Rich media taking over the web is a very positive, macro trend. Videos have very similar attributes to images and working with videos is really starting to pick up as well. Filestack lets you transcode video via API.
What are some cool examples of companies using the Filestack API?
One example we love is Pingboard. They’re building company directory software to help businesses manage employee information. They allow every employee to create a profile of themselves when they start at a company.
One of the first things that you do is upload a profile picture. They use our facial detection software, so if you’re uploading a picture of yourself with this huge canyon in the background, it knows to grab your face and automatically resize it properly.
They use our picker to grab the image, apply the blur effect, and then automatically round and resize it. By fetching from our API, it’s responsive at different sizes and on mobile.
Edtech is a big vertical for us. Many websites or mobile apps that have a profile image is a good fit. There are lots of photo sharing examples as well. Really, any business that wants to integrate their users’ content is a good candidate to work with us.
How is the business growing?
We just passed $2 million in annual recurring revenue and that number is growing 100% a year now. It’s accelerating. We handle nearly 1 billion files every month and have processed more than 20 billion total.
Over 50,000 developers have at least experimented with our service. If you check out our website, we have a pretty generous free plan. We get anywhere from 50 to 75 new signups a day. Then, we work hard to figure out how to get paid conversions out of that, when developers really build us into their websites and mobile apps.
It’s been a challenge to reposition and rebrand the company but the benefits are worth it. Our customers are realizing more and more what we can offer. Our sales people are pitching the entire product, not just the picker.
The brand really got everyone seeing and believing that we do more. File uploading is still the core of what we do but it’s not the only thing we do
What marketing experiments have you run?
We’ve experimented with the basics like pay-per-click and Facebook ads. One of the challenges with being new is that it’s difficult to gauge what has worked and what hasn’t. We’re very slowly ramping things up and making sure we have the proper metrics in place.
We’re improving our content marketing. We have pretty robust documentation and that drives a ton of traffic. Both developers and Google seem to love it.
Also, our product itself helps with marketing. Our file picker and file viewer (for converting a PowerPoint or any file to a web page preview) are installed all over the web and often have a tiny little bit of branding on it. Lots of our users see this and might decide to try out Filestack. By the way, you can pay a little extra to remove the branding.
Tell me more about the process of renaming the company.
We had a niche name that only lent itself to one aspect of our product, granted, an important one. Some companies can get over this like Salesforce, but we couldn’t escape it. ZenPayroll changed their name to Gusto right before we did this and internally we considered their approach. I used it as an inspirational example. In my mind, it’s never too late to change your name if the future is bigger than the past.
I’m a fan of more descriptive names. I’m staring at the white board where we brainstormed 50 names. Eventually, Filestack stuck and we were able to get the domain name for a reasonable price.
We ran the two sites side-by-side initially and all new customers would go to Filestack. This let us deprecate old endpoints and better design the API subdomains. Admittedly, we’re still phasing out some of the old Filepicker branding elements but we’re making progress fast.
What stack do you run?
The original technology was all python and powered the company for 4 years. Our picker is powered by an Angular JavaScript library.
We’re now doing everything in Go and finally have automation to auto-scale our infrastructure. We transitioned off of Mongo because we didn’t have good insight into performance and we’re now using a PostgreSQL database. We’ve got real tight integration with Fastly for content delivery .
Before we were powered by ImageMagick to do a lot of our transformations. We spent about three months rebuilding this and deployed a Go-powered processing engine that uses a bunch of different tools. ImageMagick is still a part of it, but it’s a much better framework. This powers facial detection and the viewer.
We’re constantly assessing what we should build on our own and who we should partner with. CDN through Fastly was a no-brainer for us, but there could be other interesting products that do object detection and other machine learning tasks we could build into the platform.
File manipulation is rarely a product’s core use case, but it’s often a helpful feature for any app with user generated content. Filestack is a great example of how developers can gain leverage by outsourcing these features to 3rd-party APIs. Thanks to Pat, Dylan, and the Filestack team for sharing the inside story on the company’s growth and new name!
Want API interviews in your inbox? Subscribe now.
Also, check out this API Starter Pack for a free API bundle with $10k+ in credits.
Kin Lane is a big voice in API evangelism. He’s currently going through some personal struggles and you can support him here.
