Conversation with Keibi: They Protect Social Networks

I had the chance to meet with the executive team from Keibi yesterday in San Francisco. Keibi has created a way to help social networks find and remove objectionable content quickly and easily. The company was founded a year ago, has raised $5M in VC funding and the name comes from the Japanese word for safety/protection.

The idea is simple: social networks get inundated with content that is outside of their terms of service. Typically large social networks hire cheap labor around the world to sit in front of a computer and watch images and text coming in for those pieces that are violators. Keibi uses their matching technology in their "Keibi Moderation Suite" to find images that might be objectionable and automatically can either flag for follow-up or delete. For example, if an image has been deleted once, Keibi can check all future images for dups and automatically delete them. This technology can dramatically save costly labor for a social networking service. The social network can hook into Keibi’s API in minutes or can use a spider method. They suggest that the API is the more popular method.

Piczo is their first public client (they have several other private clients) and they shared the following quote from Keith Crowell, Director of Member Services and Safety at Piczo. "Since we started using Keibi, we are now reviewing more than 200 times the images we used to, while spending 70 percent less on related overhead."

This software can also help advertisers become more comfortable with advertising on social networking sites by verfiying that their brands never appear near potential TOS violations. Keibi runs as a service (aka ASP model) and is priced based on usage (they wouldn’t share any figures).

Image verification is the first release of Keibi. Their image recognitiion software looks at all of the different signals including flagging and past user performance. It also checks the friends of the inappropriate party as their research shows that their friends probably post inappropriate content as well – "where there is smoke there is usually fire".

Next up for Keibi is text matching. I noted that this could be a great spam killer and they responded that it will be a great spam killer but also a scam finder. I think it would be great if they offered a plugin for forums and blog commenting. I could see a very large market opportunity for Keibi as social networks (and their exploits) grow in size.