Murdoch Eavesdrops On Your Chats To Give Recommendations

Murdoch wants to give you what you didn’t ask for, yet.

Instant messaging is overshadowing social media, and that’s creating new opportunities for data mining and delivering useful content. Murdoch has built a way to listen in to your instant messaging conversations and chime in with recommendations for you and your friends. Today at TechCrunch Disrupt London’s Hackathon, a scrappy team of developers revealed the MurdochBot for chat app Telegram.

And yes, the name seems to be inspired by Rupert Murdoch and his voicemail hacking adventures.

When you add “MurdochBot” to one of your conversations on Telegram, it will first ask you a few questions about your favorite television shows. Murdoch then parses this data using IBM’s sentiment analysis Alchemy API and Sky TV’s archive of viewership data to determine your style. Then when the time is right, Murdoch can speak up and offer suggestions. Right now it focuses on recommending television shows, but the technology could be expanded to point people to restaurants, apps, and more.


By building the interface directly into a popularly used chat app, you don’t have to download something separate, or even keep remembering to open it. “These instant messaging apps are where the next app ecosystems will be formed” says Ben Dixon, one of Murdoch’s developers.

Murdoch joins a hot market where tech giants are also trying to combine messaging with artificial intelligence assistants. Facebook’s M sees users communicating directly with the AI. But that means you have to purposefully ping it whenever you want help. The problem is you might not know what M can do or when it would be useful.

Dixon says “If you ask people what they want, their answers don’t tell you very much, so you learn a lot more by listening to what they say naturally. Passive data like that is much more accurate in predicting what people want.”

Screen Shot 2015-12-06 at 8.44.59 AM

Him and fellow Murdoch developer J. White will keep Murdoch available on Telegram while continuing their gigs at on-demand staffing startup Catapult. Sky TV has also expressed interest in playing with the app.

It will take users some time to adapt to AI personal assistants. But there’s immense potential for them to conveniently help us out, especially if built into chat apps. This is where we’re spending time now, so Dixon concludes that, “companies can become part of that experience and improve it, rather than trying to control it. Messaging apps are going to be much more about becoming part of the conversation.”