Improving the Web’s Fabric – an Interview with Paul Doyle of Fabric Technologies

Fabric Technologies is a startup in Montreal that recently raised financing from Real Ventures. They were Real Ventures’ first investment. The company was founded by Phil Taylor (CTO), Paul Doyle (CEO) and Peter Zion (VP Engineering). They all worked together at Softimage (originally part of Avid, acquired by Autodesk in 2008) as part of the XSI team. The company now has five people. Phil initiated the project in late 2009 and brought on Paul and Peter in the summer of 2010. We spoke with Paul to learn more about Fabric Technologies and their product Fabric Engine.

NextMontreal: What is Fabric Technologies?

Paul: We’re a Montreal based startup with a strong background in high-end 3D. We’re working on a product called Fabric Engine. It’s a high-performance multi-threading engine that we can use in a few different ways. One of the coolest areas is using Fabric Engine as part of the browser so web applications can use the full power of the available hardware.

NextMontreal: The current business is a significant pivot from where you were originally focused in the 3D market, right? What made you change focus? What opportunities do you see in the market that are driving you to create Fabric Technologies?

Paul: It would be nice to say there was a ‘Eureka!’ moment and we were suddenly ‘high performance computing in the browser’, but it was actually a slow process. Our original plan was to go after the opportunity we saw in the 3D market – we were building a multi-threaded customization framework that used web technologies to build the tools. We were getting great responses from the studios we went to see, and we could have built that business up. It’s still a market that Fabric applies to, we just aren’t focusing on it in the same way.

The first change came when we saw we could put Fabric completely inside the browser – that we could bring high-end interactive 3D performance to websites and web applications. The second change happened when we saw that the framework we were building was not 3D specific, it was actually not specialized at all – Fabric Engine can handle any computationally intensive task. When we sat down as a team and talked it through, it was clear that ‘a browser-based multi-threading engine for the web and mobile market’ was a pretty compelling proposition. We were excited by it, and so were the investors.

As for opportunities – we see two technology trends that are in conflict. The browser is becoming the hub for most of our computing, the trend is for that to continue. At the same time, hardware is trending towards multiple cores and hybrid (CPU and GPU) architectures. The browser can’t use most of this stuff – there are bits and pieces that get accelerated, but mostly it’s single-threaded and running an interpreted language. It’s just not going to be as fast as a multi-threaded compiled application. If you can enable a browser to access those resources, then you’re looking at web applications that can handle heavy computing tasks. Lastly, a Fabric application will run on any browser that has the Fabric plugin installed/embedded – the bane of developers is building apps for multiple platforms. Fabric addresses that pain without having to swallow a drop in performance.

NextMontreal: So users will need the Fabric plugin in their browser to run apps built using Fabric?

Paul: Yes – Fabric for browsers is installed as a plugin. Native Client will mean that will eventually not be the case on Chrome and Chrome OS (any any other browser that uses NaCl) – Fabric will just run. We are working to partner with browser and device vendors to get Fabric embedded so we can remove the plugin install friction. We are also looking at some cool things we can do with ‘virtualized’ Fabric Engine – one of which would be to allow vanilla browser clients to have Fabric applications running in them without the plugin installed.

NextMontreal: What was the fundraising process like? How long did it take? What did you learn going through this process (I believe it was the first time for you guys, right?)

Paul: It was the first time for Peter and me, Phil had raised money for his first company, CAT, which was acquired by Avid in 2006 in NZ without any problems at all.

It was pretty stressful at times It took us six months to close with Real – I was fortunate to meet Mark MacLeod early on, as I was looking for CFO advice. As we talked things through, it became clear that he liked the team and thought we could be a good fit for Real. We kept in touch over the next few months, and as soon as Real closed their fund in October we went and presented to the other partners. That went well and we got everything closed in time for Christmas (it sounds so easy…).

At one point I was juggling about 20 different investment threads – plus all the emails and calls trying to get in the door elsewhere, plus working on validation and strategic partnerships. At the same time we were pivoting and changing so often that I eventually gave up on the business plan completely – it was usually redundant before I had finished a new iteration. A simple pitch deck and a demo is all that we needed for the investor meetings. That said – the exercise of writing the plan was useful at the start. You need to know those things, even if it’s just to discard them later. We still write a lot of things down and share them on google docs – assuming everyone has the same model in their head is dangerous.

Key learnings:

• get a lawyer who understands startups. If they understand startups then they might agree to bill you upon funding. We got introductions and all kinds of help from our guy. When it came to the term sheet, it was clear how valuable a lawyer is.
• at every stage, it takes longer than you expect
• if you have co-founders, then have them directly involved in negotiation. I was handling it on my own and it caused problems – if you aren’t able to make unilateral decisions, then you need all the decision makers there. As soon as we fixed that, it became a lot easier.
• some investors will ignore you, some will say no without having listened to you, some will get excited but want you to do something else. Don’t take it personally, listen to everything.
• use your network. Be shameless in seeking introductions.
• read everything you can and get good advice.

NextMontreal: Who is the target customer?

Paul: Developers of web and mobile applications. Developers of native applications that want a route to get their applications in a browser. Developers that want an exceptional engine for handling multi-threading (Fabric Engine is not limited to the browser).

NextMontreal: What are the ideal applications for the technology?

Paul: Initially we see the big push is going to be on taking web applications from being data intensive and seeing them become compute intensive as well – this can mean an improvement in the performance of existing web apps, but also a movement of desktop apps to the browser. As Fabric gains ground we expect to see websites using our technology for content as well. One of the most important things about the Fabric plugin is that it works as part of the DOM – it isn’t a blackbox being hosted inside the web page, it’s a part of the web page. Everything about Fabric is web developer friendly – this is critical if we are going to succeed. So things like the interface for a Fabric application are all built with vanilla HTML and JavaScript, Fabric just handles the heavy lifting.

NextMontreal: When will you be putting your product into people’s hands?

Paul: We will be starting our closed alpha in the next few months. If things go well then we’d like to start our beta in the summer.

NextMontreal: What’s the competitive landscape look like?

Paul: It’s a busy space for sure, but we haven’t seen anyone else taking our approach.

There are some great tools for software engineers to write better code for multi-core architectures, but they don’t help in the browser and they are pretty hardcore engineering tools.

There is WebGL which is all about the GPU, but there’s nothing there for CPU compute (or blending the two) – it’s also very new tech.

There are the new libraries for Flash, but again they are GPU focused and Flash has its own problems.

Native Client from Google is very cool, but we see NaCL as a complement to Fabric – it’s a good way for getting C++ apps hosted inside the browser, but it isn’t part of the browser and web developers aren’t going to start writing C++ if they can help it.

There are a lot of solutions appearing for ‘build once, deploy everywhere’ for mobile devices – but they aren’t going after high-performance.

NextMontreal: What’s the focus for Fabric Technologies over the next 6 months?

Paul: We’re working hard on the core technology and the developer tools so that we can get to closed alpha and start working with customers.

NextMontreal: What have you learned as startup founders so far?

Paul: I can’t speak for the other two, so this is just my viewpoint.

• working with the smartest engineers you can find is a the smartest thing you can do.
• however long you think something will take, it will take longer.
• we should have been more cautious in who we approached to be on the team – I had a stressful few days getting releases signed by people that had been briefly involved with Fabric. They weren’t bad people (thankfully they all signed) – just not committed to Fabric. I should have established that commitment was there before we agreed to work with them. Thankfully we didn’t burn any bridges.
• you never know who you might run into again someday – it pays to be polite. You never know when someone is in a position where they can affect your future.
• there are so many blogs, websites, podcasts and books that can help you. Be constantly open to learning and you will benefit greatly.
• watching The Social Network should be something you make your co-founders do before handing out board resolutions for them to sign

Related posts:

1. Real Ventures Announces First Two Investments – Fabric Technologies and DokDok
2. Teledyne Technologies to Acquire DALSA Corporation
3. Parker Hannifin Acquires Quebec-Based Micro Thermo Technologies from Carrier
4. Google’s App Engine, Cloud Computing and Platform as a Service – an Interview with Sean Lynch
5. Crazy Cool Toys, Tech and Gadgets – an Interview with Davin Sufer from WowWee