What does Speedi.ly do? One thing, very well and at scale. Speedi.ly takes a piece of content, or grabs the content from a URL, and analyzes it. It does this very fast and it outputs some key data. Speedi.ly tells you the language of the content, categorizes it (topics, keywords), and additional metadata. This metadata payload is exactly what Robert Scoble is talking about with his SuperTweet idea.
Here's what Speedi.ly returns for this story we wrote on the Skype/eBay sale:
Speedi.ly successfully categorizes the story as about technology. Not bad for on the fly and human-free categorization. You'll see the field for entities as well, which is currently blank. Speedi.ly will soon turn that on as well.
Now check out the results for this article from the SF Chronicle, properly categorizing it under sports:
Why is this useful? Most URLs passed around today on Twitter and Facebook are completely metadata free. Search engines like Topsy are forced to look at the text in the Tweet or status message, if any, for context on what the URL is about.
Even Digg and Delicious rely on data entered in by humans to categorize URLs. With a service like Speedi.ly, those services can create a sort of real time page rank on the fly.
If you want to try it out yourself, go to http://classify.speedi.ly/fun and us login:customer and password: logmein. Note that this isn't going to be a huge wow moment for most users, but potential partners will be able to see what Speedi.ly is capable of.
There are a couple of other services sniffing around the same space as Speedi.ly. Factery, which we wrote about earlier this week, looks at shared URLs and pulls out key facts. And Thompson Reuters has OpenCalais, which has 18,000 customers.
I interviewed Keith this afternoon about the service on video. It's embedded below. He also mentions that Speedi.ly is already working with a partner to categorize URLs on the fly.