The Sosh app provides a ‘social concierge’ service, which suggests things to do depending on where you are, the time of day, your interests and the known interests of your different sets of friends.
There are a number of apps in existence such as Google Now, Donna, Tempo, Sara and EasilyDo that are designed to provide personal assistant services which try to anticipate users’ needs. However, they tend to start out from a work-and-timetable perspective rather than focusing on a person’s own interests. Instead of mimicking the Foursquare and Yelp approach, where people are looking for places – restaurants, stores, cinemas etc. – Sosh, developed by US startup Offline Labs, suggests places to go according to what it has learned over time about each member. Sosh aims to offer a personalized ‘concierge’ service on mobile to help people find interesting places to go and things to do. The app is already up and running in San Francisco and New York and has just been launched in Seattle. This ‘social concierge’ concept seems to be enticing investors, among them California-based venture capital firm Khosla Ventures, which last summer contributed to a $10 million Series B funding round, bringing the total invested in the firm to over $16 million.
Recommendations based on the interests of members and their friends
The app analyzes and aggregates a large amount of data on every event, place and activity in the city where an app subscriber lives. It also gathers around 100 ‘explicit’ data points on each user’s preferences, plus 20 times this number of ‘implicit’ data points, i.e. background personal details gleaned on the basis of explicit data, but which is not provided intentionally. Sosh then uses machine learning techniques to link up events, places and activities with app users, according to their interests and those of their various sets of friends, and makes tailored recommendations based on the time and current location of the subscriber. For example, if a user opens Sosh on a Friday evening, in a restaurant, the app will certainly suggest a bar or a concert nearby. The app tracks people’s social graphs to get to know the interests they have in common with particular friends, and can constantly learn about the kinds of things they like to do, singly or together. The app’s ‘social’ functionality also allows users to share the suggestions easily on social networks such as Facebook and Twitter.
Recommendations involving the social network
Offline Labs co-founder and CEO Rishi Mandal points out how the Soshi app is designed to enable the automated recommendation service to develop. As Sosh gradually gets to know the type of activity that each member and his/her friends enjoy and which people or sets s/he would tend to go with to a given event or activity, the app will be able to make more and more accurate suggestions, sending such notifications as: “There’s a special event in such-and-such a bar tonight and you and Mathieu are both free this evening. Shall I book you an Uber [rideshare] so that you can go this evening?” Sosh still has a long way to go to build out this kind of predictive recommendations program. However, the team has already gathered a great deal of data, which means it has something to offer companies who wish to get to know more about their (potential) customers. Sosh does not currently charge these companies, but the data flows could prove a money-spinner. Offline Labs is planning to expand Sosh to a number of other US cities over the next year, starting (probably) with Chicago, Los Angeles and Boston.