Portugal Ventures-backed company Wizi knows a thing or two about social geolocation services, having made a business in recent years building Foursquare-like apps for carriers, including Telefonica and Vodafone. One example being Vodafone Radar — an app which it says garnered around half a million downloads.
But working with carriers, say co-founders Paulo Dimas and Jorge Medeira, is painstakingly slow compared to the pace of startup development — taking, for instance, a full two years to launch just one app with Telefonica in Brazil. (“It’s horrible” is their concise verdict on working with carriers).
Which is why they decided to branch out on their own last April, launching a social location app for checkins and keeping abreast of friends’ whereabouts called FacesIn, initially for Android. That pretty standard ‘SoLoMo’ service has now morphed — via a partial pivot — into something more interesting: delivering location-based alerts on where people of interest to you will be in the near future.
So rather than continuing to chase the bandwagon of apps and services that promise to keep you appraised of your friends’ real-time movements, such as the aforementioned Foursquare or Swarm or other real-time location mapping/pinging services, FacesIn can now parse public data and social postings to alert you when someone you are interested in is going to be at a public event in your vicinity soon — such as speaking at a conference. The idea being to give users the chance to cross paths with people they want to network with.
“Everybody is doing ‘the now’ — so who’s nearby now. That’s interesting. But nobody is doing ‘soon’. Nobody is doing who is going to be in this location in the near future,” says Dimas. “So we believe there’s a lot of value here. For instance if you are in London and you want to pick events to go to, typically people do it because they want to meet somebody, they want to connect, they want to follow up.”
If you were talking about pro-actively monitoring the schedules of celebrities or general web users that would sound pretty stalkerish but the initial focus for the new version of the app is on tech startup entrepreneurs and investors. Who of course have a special thirst for networking, contacts building and co-locating.
Using online data sources such as CrunchBase and AngelList, the team has curated a list of around 500 “VIP” individuals of especial relevance to this group — including TC’s very own Editor-At-Large, Mike Butcher (I checked and they confirmed Mike is on their list) — cross referencing this list with public information sources such as Eventbrite and Twitter to gather data on these people’s future locations so it can keep its own users in the loop.
It’s also pulling follower and other data from AngelList and Twitter to get info on who its users are most interested in (such as people they follow and actively engage with on these social platforms); and from LinkedIn — again to get info on who users may be most interested in meeting (via their contacts list). FacesIn does not scrape VIP targets’ location data from LinkedIn itself; only public information sources are used for gathering their location signals.
“We try to infer the relevancy. That’s a big challenge,” Dimas tells TechCrunch, discussing how it determines when it should send a notification of a tech VIP’s future location. “We are trying to overcome major challenges in this domain… We try to make it as automatic as possible to the user. Because we know users are lazy and so we try to, by combining the distance to the location, the relevancy — the affinity level (we have a kind of a score for each contact that you have, or each person that you follow), and by combining these variables we decide if this notification is relevant. Because of course we don’t want to spam the user. And we also don’t want to ask many questions to the user. That’s why you turn on LinkedIn and it does the work for you.”
Even if they weren’t able to plug into LinkedIn — say if the platform decided to close off its API to them — the team says it could still use smartphone contacts data to power its service (although it stresses it has a good working relationship with LinkedIn). So it’s not wholly dependent on any one of these other platforms, even though it does rely on public data to power its predictions.
What about business model? Wizi isn’t expecting tech folk to pay to be pinged that Butcher (or another startup luminary) will be passing through their city in a few weeks. The app is entirely free for now. Rather, in future, the co-founders believe there’s a business to be had in selling its social utility to salespeople who also have a special need to coincide in meatspace with potential leads so they can press flesh and seal deals.
But first they’re testing and honing their concept on the startup crowd. Because these sorts of predictive notifications need careful tuning. There is obviously a fine line between utility and spam — given the app needs to determine which of the people who you find interesting are you really most interested in meeting.
How does it do that? Natural language processing and textual analytics of your social activity, coupled with machine learning algorithms is the team’s special sauce here. To help its algorithms improve, the app asks users to give feedback on whether individual notifications were useful or not so it can better learn their preferences.
FacesIn is also designed to be an “invisible” app — so it runs in the background and only intrudes on the user when it has something to tell them. The team say they are aiming for a maximum of two notifications per day.
“What we are doing differently is talking about the future, and not about the past and the present… Of course it’s a big challenge. If it wasn’t everybody would be doing it,” says Dimas. “We have now combined a set of technologies — namely in the domain of natural language processing, and also machine learning, and that allows us to interpret, to understand when for instance somebody is going to be a speaker at an event. We collect that data and combine it with the profile data that we already have from the social networks.”
“We are doing this progressively,” he adds. “What we are doing to start and to gain an initial critical mass is to focus on a specific target of users. In this case we are targeting the startup ecosystem… And so by analyzing all event data that are published by CrunchBase… we have lots of data that we can process.
“Of course people may say they are going to be at an event, or they are announced as speakers at the event, and they may not be there in the end. But we just link them to the event page to keep the user informed about all the details.”
People whose locations users of FacesIn are monitoring are not informed their movements are being tracked and re-broadcast in this way — the app has no public list equivalent to Twitter’s ‘following’ list. However the team stresses all the future location information it is drawing on is public data; so it’s just joining a few dots and automating notifications. Albeit that may still feel a little creepy. But this is what you get when dispersed pieces of (big) public data are brought together with a little lightweight data processing.
“We believe there is so much value in this but when you talk about it… the reaction is ‘oh, isn’t that freaky?’ But it’s possible to do. And it’s valuable for the user,” says Dimas. “Aren’t we in a world where privacy is always a concern with everything you do?”
The team has pulled in around €500,000 in seed funding specifically for FacesIn from Portugal Ventures. It’s looking to raise a bridge round in the next 12 to 18 months to continue developing the product and — it hopes — turn a free social utility that’s pitched at the startup ecosystem into a service that salespeople will be willing to pay for. An iOS version of FacesIn is also planned for Q2 or Q3 this year.
Asked about competitors they name-check a few social location apps — such as Highlight, Glympse and Connect — but again point out that rivals are solely focused on real-time location (as FacesIn was prior to its pivot), whereas their new flagship feature is all about looking ahead to the people and opportunities that are headed your way.
Now the team needs to prove it can deliver genuinely useful predictive signals — rather than generating yet another feed of location-based noise. The proof of this algorithm’s utility will be in the selective deliverance.