Grokr, the Sun­ny­vale-based mobile start­up, is poised to unveil a brand new iOS app which has the poten­tial to rev­o­lu­tion­ize the way search gets done on mobile devices.

Sim­i­lar to “Google Now” (the Android app devel­oped by Google to offer proac­tive results based on pre­vi­ous search behav­ior), Grokr has some addi­tion­al tricks in its bag which should intrigue mobile adver­tis­ing agen­cies far and wide. With search tech­nol­o­gy like this com­ing on stream, new mobile adver­tis­ing strate­gies may also emerge.

Google Now plus extras for iOS

While Grokr can do pret­ty well every­thing that Google Now can, like pro­vid­ing updates on every­thing from the weath­er to your favorite sports team to the traf­fic, it also sucks in data from your social media pro­files on LinkedIn, Twit­ter or Face­book (Google Now con­fines its results to its own ser­vices). The start-up’s CEO and co-founder, Sri­vats Sam­path, explained that he want­ed to rein­vent search on mobile, mov­ing away from the tra­di­tion­al approach of sim­ply minia­tur­iz­ing search meth­ods designed for PCs.

Instead, Grokr has shift­ed from a key­word-based mod­el to one based on the mean­ings behind the words, and from a man­u­al process to a pre­dic­tive one: it can dis­tin­guish between Taj Mahal the singer, the hotel and the mon­u­ment, for exam­ple, based on user search behav­ior. It draws not only from Free­base, the com­mu­ni­ty-con­tributed data­s­tore used by Google, but between 50 and 75 oth­er data sources like Rovi, Fac­tu­al and Yelp. The results?  As Sam­path puts it:

“Right now, we have 700 mil­lion facts in the data­base, and we have 25 mil­lion enti­ties. Then we built machine-learn­ing algo­rithms around that. That allowed us to rank those entities.”

A new pre­dic­tive search engine?

Although its Google Now-type proac­tive alerts will entice many, Grokr can also be used as an alter­na­tive search engine. The search box in the app gen­er­ates web-sourced results based on Grokr’s under­stand­ing of its users’ interests.

Sam­path continued:

“We’ve essen­tial­ly built Google Now for iOS, but what we’ve also done is added more sig­nals than what Google Now has. Google Now pri­mar­i­ly uses loca­tion and time of day. We use loca­tion, time of day, your inter­ests, your intent – just a whole bunch of sig­nals to bring forth all this infor­ma­tion to you.”

And there’s no track­ing of indi­vid­u­als involved – all data is per­son­al­ized and anonymized.