X
Tech

Siri acquired by Apple; iPhone <em>becomes</em> the Virtual Personal Assistant?

Apple buys Siri, and brings real semantic smarts to Cupertino.
Written by Paul Miller, Contributor

According to an FTC filing unearthed by Robert Scoble earlier this week, Apple appeared to have acquired Siri. The rumour has since been confirmed, and I spoke with Siri board member Norman Winarsky to get some insight as to what this news might mean for the semantic technology startup I've been following since before it emerged from stealth back in 2008.

Siri reappeared on the scene back in February, launching their innovative (US-only) iPhone app after a protracted silence. In March, the company snatched Gummi Hafsteinsson from Google, strengthening their mobile credentials still further and appearing to maintain momentum. Semantic technology enthusiasts watched for Siri to get even smarter, mobile officianados waited with bated breath for the promised Android app, and those of us outside the United States waited for a version of the app that understood our accents, our geographies, and the APIs of the companies offering travel, events, entertainment, restaurants and the rest in our territories. Although Winarsky, VP for Ventures, Licensing and Strategic Programs at the company from which Siri was originally spun out, felt unable to comment on the specifics of the product road map moving forward, I can't help speculating that any Android (or Blackberry) app has probably moved a lot further down the priority list.

Even with billions burning a hole in his pocket, it's unlikely (although not impossible) that Steve Jobs spent more than $200 million just to interfere with the release of an Android app. So what did Apple want?

Firstly, they probably wanted to make sure that someone else didn't pounce and grab the opportunity. Secondly, Siri is smart, innovative and impressive. The application does a great job of appearing simple and intuitive, whilst fulfilling a useful - and potentially complex - role. Given the surprisingly limited voice capabilities of Apple's current mobile devices, they may well be interested in harnessing some of that language interpretation capability for future mobile devices and even their desktop line. I doubt that required Siri, though, so much as the far cheaper poaching of some key staff from Siri or their partner (and fellow SRI spin-out), Nuance.

In its current form, Siri remains relatively limited in scope (finding and booking things on the move, essentially) and geography (the United States), and if Apple is serious about making something of the company they've acquired (rather than just securing its staff and IP) it will need to move forward aggressively on both fronts. Do they want a travel assistant with global reach, or do they want to give every iPhone/iPad/Mac owner their very own virtual secretary?

Winarsky is quick to suggest that the notion of the 'virtual personal assistant' applies in contexts far beyond the mobile niche carved out by Siri. Citing a 'perfect storm' created by the conjunction of bandwidth, compelling user interfaces and on-demand network compute power, he argues that Siri's current incarnation merely represents a taste of what's still to come. SRI continues to build upon the CALO project from which Siri was born, and opportunities are becoming increasingly clear in verticals as diverse as healthcare, retail, and call centre operations. SRI conducts 'more than 2,000 projects per year,' and the next CALO spin-off could emerge from their labs 'in less than six months.' Surely Apple didn't buy the wrong CALO baby?

Far from embarking upon a mission to solve the Turing Test, which Winarsky argues cannot be done, SRI researchers (and the Siri team) are amongst those transforming the tarnished image of Artificial Intelligence. By working within a defined vertical (healthcare, etc), it is entirely feasible to train an agent in such a way that it can reliably infer and act intelligently with respect to context. When you're only dealing with medical complaints (say), it's not impossible to train a machine to excel with idiom, jargon, and even ambiguity. It's also feasible to expect that the machine could learn - and adapt - 'in the wild,' without recourse to the lab every time external factors shift substantially.

As well as pointing to SRI's long heritage, successful track record, and future plans, Winarsky was keen to stress the important role played by Siri's venture capital backers in reaching the current settlement with Apple. Both Menlo Ventures' Shawn Carolan and Morgenthaler's Gary Morgenthaler apparently 'did the heavy lifting' on the deal. Winarsky points out that the hardest part of getting a company like Siri to this stage isn't actually developing the technology, but recognising the real market opportunity and 'navigating amongst the giants' in the space. In all of this, he reiterated, good VCs proved invaluable. And they weren't even listening in on the call!

I never thought I'd see 'metadata' discussed in the UK Parliament, then I did when we adopted Government Metadata Standards a few years back. I never thought I'd see 'RDF' pass the lips of a head of state, then Prime Minister Gordon Brown said it this year. Will Steve Jobs stand up at WWDC'10 and talk about Apple's embracing of/invention of 'the Semantic Web?' We shall see!

Good luck to all at Siri, and I look forward to seeing what Apple does next... and what SRI spawns next.

Editorial standards