The idea behind Gaston Labs is to come up with ideas that seem reasonable and see if the market agrees. Our methods are inspired by the Lean Startup methodology (Eric Ries is a friend and an investor in Gaston Labs). This involves coming up with ideas, vetting them, and then, if needed, firing them. Today, we fired our first idea and in the spirit of airing postmortems, we thought we’d share our thinking. (Aaron Patzer wins the prize for burying Swift along with a 25-page epilogue.)

The first idea that floated up to the top of our pile was to create the world’s first iWork-to-PDF service. If you go to nerd events/conferences, you’ll see nearly everyone using Keynote or Prezi; pretty much nobody in the early-adopter segment uses PowerPoint. Any web service that accepts uploads should know what to do with an uploaded Keynote file but we noted that none of the major services knew what the heck to do with a Keynote file: Dropbox, Box.net, Google, Microsoft Live Drive, Sharefile, YouSendIt, Scribd, or even SlideShare.

So we developed ConvertAllTheThings, a web-only API that takes a Keynote file and spits back out a PDF. While Zencoder does a similar thing with video files, as far as we know, we have the only service to do this on the Internet. The name was halfway between silly and terrible and way too long, but frankly it wasn’t likely to matter as this wouldn’t be a consumer service; it would strictly be B2B, like Zencoder. Here’s a cool demo we built or a video of the demo in use. We were pretty psyched that we could put together a novel and valuable service.

In parallel, we started reaching out to potential (large) customers. Unfortunately, most of them wanted a “one stop shop” provider that could transcode all all their documents, not just iWork. Unfortunately, this would have required us to transcode Microsoft Office files. Having heard some of the inside track of what Crocodoc had gone through in getting properly licensed by Microsoft was a huge turnoff. And while we have previously built a transcoding pipeline based on OpenOffice (as have most of the above vendors), the quality is definitely a few notches lower than what Office proper can put out.

To put a nail in the coffin, our clients asked for the technology to be onsite, since they didn’t want their customers’ data leaving their datacenters. The technology stack we had put together for CATT was focused on OS/X, which is difficult to virtualize for both technical and licensing reasons.

While all of these roadblocks are circumventable, it seemed a good point at which to pause, take a step back, and ask ourselves if the market was likely to be fruitful. The initial concept of having a quickly-deployable “SaaS for SaaS” business that could operate with decent margins on a  simple Twilio-like pay-per-use model and lights-out cloud management seemed really appealing, but the reality of the market seemed more like an opening for an enterprise Microsoft transcoding consultancy (ew) with a potential market size of low millions per year in the best case, after a huge amount of work and ongoing support. So we’ve decided to take a pause on this, leave the demo live, and move on to the next idea for now.