Cutting Your Losses Early

My little pet-project (lifestreaming service) is hereby declared dead. Over the last months, I have put effort in rewriting the core architecture of Kakuteru, the activity stream aggregation framework, but I’m aborting it now for several reasons.

To my dear alpha users: I sincerely apologize and thank you for your support over the last quarters. I will continue to run the service for another month and you can download dumps of your data. Also, I will make available all source-code and try to wrap some intelligent components into Ruby Gems.

The Real-time Web is still in it’s Infancy

With this comes great opportunity, but also great challenges. When I started Kakuteru, I made a quick mashup using Friendfeed’s API and I have complained several times about the lack of meta-data available in it’s stream. After looking around I decided to start doing my own aggregation, pubhubsub was not yet mature (and still isn’t). Gnip was a very interesting service, but it didn’t support enough different services.

Aggregating all those feeds from passive API’s is very expensive, I wrote a framework in Ruby that is pretty damn solid using Nanite (fabric of ruby agents), but you still have to deal with rate-limiting, format changes, scaling, etc.

In other words, getting a huge pipeline of activity stream updates is NOT a commodity! Not yet anyway. When looking at prices charged by services like Gnip and SuperFeedr, you can see this is obvious. Specifically looking at Gnip, it seems they’ve gone back and forth a bit between different business models.

I am very confident however, that companies like Gnip will be major players in the Real-time Web game.


One of my big personal goals is to find ways to create value by marrying the Semantic and the Real-time Web. In my daily work, as the CTO for I’m doing a lot of innovation on these fronts with a great team of superstar developers. However, we are all busting our asses, so there is not a lot of time for big undertakings like Kakuteru on the side.

However, I do intend to keep on hacking and build mashups in the wee hours. Specifically, I want to create Nano-startups that make use of this emerging ‘Synaptic Web’. So for starters, I’ve rebranded my blog to, where I will push out well-digested startup ideas, semantic/real-time hacks and general technology thinking.