It’s Saturday afternoon. You’re bored, and you want something to do. It’s a problem that everyone has at some point. Your options are to call some friends to see what’s going on, or maybe go on Google and search “things to do in San Francisco,” but that’s unreliable. Yelp is another option, but it’s hit-or-miss most of the time. Trying to solve this problem is Sosh, which has answered the “What should I do tonight?” question over two million times in the last year.

Sosh is a product of Offline Labs, a startup founded by former Google and Slide employees, which announced $1 million in funding in 2011 from a number of angels and investment firms, including Sequoia Capital and General Catalyst Partners. Offline Labs has the goal of improving the offline, physical world by harnessing the power of the online, virtual world. The first service that follows that mandate is Sosh.

The original idea for Sosh originated from the problem of not having an organized list of “things to do”. As cofounder Rishi Mandal explained to me, “All of the data was there, but there was no way to find it in a simple and enjoyable way.” Seeing the opportunity to provide a service that people are actively looking for, the team decided to tackle the problem.

The local “What should I do?” problem isn’t new, but Sosh is taking a different tack than other localized companies. Instead of providing a list of venues, Sosh provides a list of events with a focus on the here and now. According to co-founder Rishi Mandal, the goal is less to be a Yelp replacement and more to be the “personal concierge for your social life”.

For example, if you ask Sosh for something to do on a date in San Francisco, it doesn’t ask you what type of food or location you want, but instead gives you a simple display of ideal ideas. My request turned up a class for making stuffed pasta at flour + water on Wednesday, June 20th. It’s specific, relevant, and (to me) sounds like fun. More importantly for the company, I never would have come across my personalized recommendation using Google.

Sosh learns the interests of users via an integration with the Facebook platform. With this basic demographic data, the service then asks the user a number of basic questions. What areas of San Francisco do you like? What type of events to do you like attending? Then it asks you to bookmark things you’d like to do on the website. Everything you do on the site or in the mobile app is tracked and aggregated to tailor recommendations.

Taking this recommendation data, Sosh then compares it to the existing database its compiled of what’s going on in the city. Using a mixture of aggregated data, and crawling Twitter to find out what people are talking about, Sosh maintains an up-to-date system that the company continually refines and updates. The goal isn’t to provide an index of every single restaurant or bar in a city like Yelp, but instead to provide the relevant results that people will actually follow through on.

Currently, Sosh is only available in San Francisco, but that hasn’t stunted the company’s growth very much. In fact, the company shared a number of data points that suggest the company couldn’t be more popular with its users. In the 11 months since the service launched, the average user comes back at least every two weeks, pushing the company to the point where it has now made 2 million recommendations to users.

The company has also reached a sort of critical mass within the city, as well. When the average new user signs up for the service, they already have nine friends connected via Facebook. This provides an immediate association for new users that makes the service much stickier.

As an anecdote for its popularity, Mandal said that the company provides a free t-shirt to any user that invites a friend. Initially, the company printed out 200, expecting them to go slowly. Instead, the company has now had to print thousands, with some people being so in love with the service that they try to trick the company into providing extra, free t-shirts with fake email addresses.

With such a dedicated userbase, it seems like making money would be simple. After all, there’s nothing advertisers like more than a userbase that’s willing to spend money and is dedicated to doing so. However, the company isn’t planning to try to make money off of referrals or advertisements.

Instead, the company is looking to make money by advising restaurants on how to be better. When Sosh sees that an event where a chef sits at a table with five people for a wine tasting is popular with users, the company can turn to a restaurant partner and advise them to increase the number of intimate wine tastings overseen by chefs. This advice will only carry weight as Sosh scales and reaches a certain critical mass, but from a theoretical standpoint it makes sense.

A service can only grow so big in one city, though, so the company is announcing that it is expanding the service to four additional metropolitan areas “by the end of the year”. Mandal noted that Chicago and New York are near the top of the list for user suggestions, so it wouldn’t be surprising to see Sosh roll out in either of those places later this year. However, the company didn’t confirm this.

The big reason behind the slow rollout has more to do with an obsession for perfection rather than constricted resources. The team wants to truly capture the character of a city and to understand its pulse and how it operates. This is more time-intensive, but Mandal insists that this will improve the types of recommendations that Sosh provides to users.

For now, if you’re in San Francisco, and you want to take someone out on a date and do something unique that would be impossible to find otherwise, Sosh is likely your best bet.