Post 1: Siri's Early Years: 2004 to 2010, Prior to Apple's acquisition

With all the great news about Siri and how Apple will supercharge it, I decided to share our team’s original founding vision for the Siri venture—prior to Apple’s acquisition.

Next week, in my piece: “The Future of AI Assistants.” I share my vision, post Siri, for the future of Virtual Personal Assistants. This vision helped us create several AI ventures, including TempoAI, Kasisto, Summly, and many more. 

Post 1

In 2004 when smartphones became broadly available, we were convinced that a new technological revolution was occurring. The smartphone was, in fact, not a phone. It was an always-on, always-connected supercomputer soon to be available to billions of consumers. In 2004, I wrote an article, “The Quiet Boom”, in the Red Herring magazine, on this revolution. Here are two short extracts that explain our thinking:

“The mobile phone and its accompanying wireless networks have frequently been called the computing and communications platform of the next age. Already the mobile phone has replaced the PC as the most common end-user interface and the current generation of handsets has about two orders of magnitude more power than the first PCs.

But seemingly insurmountable problems have locked up the wireless platform’s potential. Size and power constraints have limited the mobile phone to the simplest of uses, such as voice calls, games, messaging, address books and email. Those barriers will be overcome in the next two or three years as a new generation of high-value applications emerges. In the wake of the internet boom and bust, a new revolution has quietly begun.”

And this:

“There Remains a Need for “Intelligence:” Even with an effective speech interface, consumers will be frustrated if they’re required to access the services they want through a series of tedious questions. An experience as simple as buying a ticket to a ballgame and making a restaurant reservation can take dozens of keystrokes and many minutes. Instead, users must be able to easily ask for what they want, just as they might ask a real person.

We’re now seeing software agents – lightweight programs designed to perform tasks autonomously and securely – reach the level of commercialization. They can now act as “intelligent assistants” for many requests. The user can specify a request and agents will break down the tasks and reliably perform them.”

Another inspiration leading to Siri was our work on the government’s Defense Advanced Research Agency (DARPA) CALO program. CALO was a program to create a Cognitive Assistant that Learns and Organizes. This was a $150M, five-year program with the goal of building virtual assistants that could reason, learn from experience, execute tasks, be told what to do, explain what they were doing, and more. The CALO program was a great success, and we began considering how we might commercialize some of our work.

In June 2007, just after the first iPhone was released, our team began to conceive of a virtual personal assistant as an iPhone application.

Mobile phone computing and communications were orders of magnitude less powerful than today, but still, they could perform many intelligent functions. The Apple App Store hadn’t yet been created (it only started in July 2008), but it was still possible to launch an App. The dominant money-making applications were only ringtones and text messaging. In addition, each telephone carrier had its own “walled garden” and only allowed a small number of applications to be part of its ecosystem.

Over the next six months, our overall value proposition came into sharp focus: We would create a Virtual Personal Assistant App called Siri.

As an aside, many of you have asked how we came up with the name Siri. We had many names in mind before we settled on Siri. We wanted many things: no more than 4 letters, a .com domain, a name that was easy to remember, a name that didn’t have a negative connotation in any language, and so on. Finally, the CEO, Dag Kittlaus, of Norwegian descent, came up with the name Siri, which was the name of an ancient Norwegian Princess.

We worked hard to distinguish Siri from other search engines, like Google. Finally, we decided to create a virtual assistant that would not be a search engine or a conversational tool, but instead, a “do engine.” It would do things for you - not just provide an answer with links, as Google did at the time. Siri would be task-oriented, not search-oriented. We envisioned Siri to do these things for you:

• reserve something

• buy something

• remember something

• schedule a dinner & a movie: coordinate times, places

• get me specific info

So you might ask Siri questions, such as

• Find me a table for two tomorrow at Gibsons in Chicago

• Deliver me a large thin-crust pepperoni pizza with pineapple  Pizza Hut  special  for $9.99. Order it!  

• Send my wife some flowers

• Get me a 4 star hotel in San Francisco near the bay

And Siri would do it for you. The AI technology was a breakthrough developed by Adam Cheyer’s team. Speech recognition alone was not sufficient technology. We needed to translate the user request to text, understand the user’s intent, determine the apps necessary to respond to the request, fill in the “slots” in the apps, execute the actions necessary to answer the query, and respond to the user. This would be a major technological advance.

As I’ve often said in my posts: “technology alone is never sufficient to create a venture.” Siri needed a great business model. Our original business model relied on collecting fees from websites that Siri connected to. We recognized that revenue from Siri's leads to hotels, restaurants, and airlines could be substantial.

Our AI technology was limited, however. If we restricted Siri only to the market vertical of travel and entertainment, Siri would do well in understanding the user’s intent. But if we let queries go outside those areas, Siri might not always be able to recognize the intent, provide an answer, or execute the task. So, if you asked Siri, “How do I find a girlfriend?” or “Can you buy me a Nikon camera?” Siri would not yet be able to give a reasoned answer or execute the task.

We expected that as we built more capabilities into Siri, we would extend the market verticals where Siri would understand the intent and be able to act on the user’s questions. Other future market verticals would include commercial transactions—such as buying flowers, booking a taxi, food ordering, DVD rental (Netflix), and impulse commerce (books, DVDs, etc.).

We also planned more personalization:

–Remember your preferences

-Remember you favorite cuisine, flight seating preferences, events/interests

–Know your schedule

•Calendar access (Yahoo, Gmail, iCal API’s)

•Plan your Itinerary (Flight Segment & Itinerary Monitoring API’s)

–Know your friends and contacts

Here’s an example of what we imagined for Siri in 2009 as we were building Siri:

Two years after we created the Siri venture, in February 2010, we launched the Siri App in Apple’s App Store. Right away it was downloaded by many thousands of people and received great acclaim in the media. Here’s what it looked like.

Then, in April 2010, only two months after we launched the Siri App, Apple bought the Siri venture. The following year, Apple restructured Siri to serve their own strategy and vision. Then, on October 4, 2011, Steve Jobs announced Siri at Apple’s annual launch event. Since then, Siri has appeared on virtually every Apple smart device and is a major element of almost all Apple’s products.

---

Post 2 is coming next week! Please subscribe - it is always free!

Your Venture Coach,

Norman