Stealthy Bottlenose Hopes To Fulfill The Unkept Promise Of Twitter Annotations (And More)

Yesterday, at the Data 2.0 Conference in San Francisco, Twitter took the stage with DataSift to announce a new partnership to sell curated tweet data. But there’s another company that’s been working on some of the same stuff in stealth mode for over a year, and their option will be consumer facing: Bottlenose.

While the service isn’t quite ready to open its doors just yet, Bottlenose is essentially a real-time data interpretation layer on top of tweets. Well, right now it’s tweets, but eventually the plan is to open this data analysis to all types of social information — things like Facebook, Foursquare, etc. The Bottlenose team built an entirely new architecture with from scratch (an extension of the work co-founder Dominiek ter Heide had been doing for a couple years prior) to handle this data coming in. And it matters for consumers because it “totally changes the game in personalization,” co-founder Nova Spivack says.

If you’ve heard Spivack’s name before, it’s because he’s long been associated with data plays — everything from Twine to Live Matrix (which launched at Disrupt nearly a year ago). And this is the project that he views as his next big thing.

Currently, Bottlenose is a web app that looks a bit like the older version of Twitter (the view can also be changed to look more like TweetDeck with multi-panes). But it’s much more powerful. Each tweet in your stream is given a score (which is determined by many different factors including things like Klout scores, which Spivack is an investor in) and tagged with category metadata. Thanks to this, tweets can be easily sorted and filtered.

Essentially what Bottlenose has built is sort of like what Twitter Annotations were supposed to be. But Twitter has since put that project on indefinite hold.

When you send a tweet from Bottlenose, you can choose to post it as a certain type of message (metadata). This might be “news”, “humor”, “event”, “service”, etc. This helps the system learn, but Bottlenose is also doing the work behind the scenes to categorize all the tweets.

They also look at your past tweets and determine your interests (they nailed mine). This is a big part of what they serve up to you, and in that way it’s somewhat similar to what others like my6sense are doing with tweet streams.

The plan right now is to launch sometime around May in closed beta, and to the public sometime after that. While the focus will be on the web at first, there are plans for mobile apps as well. And they’d like to have a full API ready to go so that others can take advantage of the data they’re producing as well.

So what will Twitter, which has been taking tighter control of their ecosystem, think of this? Spivack notes that this is primarily a big-time data play, they just need a way to surface that data effectively with their own front-end. The team notes that they have met with Twitter and say that they have a “close working relationship” with them. Of course, as we’ve all seen, such things can change quickly.

If you hit Bottlenose’s site now, you’ll be able to request an invite for early access.