In late 2014, members of Apple Inc.'s Siri team arrived at an Amazon.com Inc. event thinking they were ahead of the competition.
Apple's three-year-old product had gained popularity for its ability to handle calendar appointments, text messaging and a few other simple tasks based on voice commands. Siri had no real competitors.
Continue Reading Below
The outlook quickly changed as the team watched Amazon's video showing off a small, voice-controlled speaker that could play music, order products and search the web. It demonstrated Amazon had figured out how to isolate voices from background noise and have a digital assistant respond to requests from a distance -- abilities Siri hadn't yet mastered.
"People at Apple's anxiety level went up a notch," said a former member of Apple's Siri team who was there that night.
Today, Apple is playing catch-up in a product category it invented, increasing worries about whether the technology giant has lost some of its innovation edge.
On Monday Apple announced HomePod, a home speaker powered by Siri that will start selling in December. The device will perform Siri functions such as dimming lights and setting reminders, though Apple touted it mainly as an advanced home-stereo system with superior audio quality. It spent years developing it.
Apple also unveiled planned upgrades to Siri for the fall that will enable it to translate English phrases into five languages and to learn from users to deliver personalized suggestions for web surfing, messaging and other apps.
Apple will enter the home-speaker market a distant third, at best. Amazon, which has been selling its Echo speaker for 2- 1/2 years, last month unveiled a device, also powered by the Alexa voice assistant, that includes a camera and a display for video calling. And it has partnered with companies to put Alexa into Ford Motor Co. cars, LG Electronics Inc. refrigerators, and General Electric Co. lamps.
Alphabet Inc., which debuted its Google Assistant-based speaker last October, said last month it will make its voice product available through an app on iPhones. Microsoft Corp. in May unveiled a speaker for its voice robot, Cortana, and Samsung Electronics Co. is expected to install its planned Bixby assistant on home appliances.
Siri has remained largely a feature of the iPhone, although it is also available on a handful of other Apple devices, including the Apple Watch. Siri's capabilities have advanced incrementally, with functions matched or exceeded by those of rival systems. Several independent studies have shown Siri is less accurate than Alexa or Google Assistant in responding to user queries.
Some former executives, close observers and even devoted customers say Apple's innovative power appears to be waning, stymied by a lack of urgency and difficulty bringing ideas to fruition. In nearly six years under Chief Executive Tim Cook, Apple's stock has soared but the company has not delivered a breakthrough product on par with the string of hits under late founder Steve Jobs, which included the iPod, iPhone and iPad.
One reason could be the iPhone itself, one of the most successful consumer products in history. It accounts for most of Apple's sales and dominates much of the company's focus, which former executives say has inhibited the company's ability to develop products untethered from the phone, as rivals did with their brand-new voice-activated devices.
"Siri is a textbook of leading on something in tech and then losing an edge despite having all the money and the talent and sitting in Silicon Valley," said Holger Mueller, a principal analyst Constellation Research, a technology research and advisory firm.
Apple's supporters say it often has entered categories after rivals and assumed a commanding position. And Apple says the pace of its innovation has only accelerated. The company has tripled research-and-development spending over the past five years to $10 billion annually. It is working on a range of projects, including an autonomous driving system, that could become hits if they come to market. Apple notes that Siri, through its presence on the iPhone, is in the pockets of hundreds of millions of users globally -- far more than any rival's voice assistant.
"We're very happy with where the company is from an innovation standpoint," Eddy Cue, the Apple senior vice president whose portfolio includes Siri, said in an interview before Monday's announcements, which also included augmented reality for developers and more-powerful Macs. "It's part of our DNA."
The race to develop digital assistants is one of the biggest areas of competition in the tech industry today. Industry executives say these products, powered by increasingly effective artificial-intelligence algorithms, are revolutionizing computing much like the PC and smartphone, leading to a future where computers will converse with humans, recall previous conversations and provide personalized responses without buttons or touchscreens.
Siri was one of Mr. Jobs's last major new products. He became a fan in 2010 when it was launched by a small startup as a digital-assistant app for iPhones. In 30 phone calls over 45 days, he persuaded its founders to sell, according to Gary Morgenthaler, a Siri investor. He then pushed them to fine-tune a handful of features that would work flawlessly across millions of iPhones in multiple languages.
Touted as "the best feature" of the iPhone 4s in 2011 by Apple marketing chief Phil Schiller, Siri converted words to text and interpreted their meaning to describe the weather or make calendar appointments and helped fuel a 73% increase in iPhone shipments in its first year. A day after the announcement, Mr. Jobs died.
In the years since, former Siri team members say, progress has been slowed by a failure to set ambitious goals, shifting strategies and a culture that prioritizes user privacy -- making it difficult to personalize and improve the product. The project also has suffered from the departures of key team members, some of whom went to competitors.
About a year after Mr. Jobs's death, Apple hired Bill Stasior, an Amazon search executive, to oversee Siri. Mr. Stasior studied artificial intelligence at Massachusetts Institute of Technology, but his expertise was in search rather than speech or language. This led some members of the Siri team to believe he didn't fully appreciate the product's original vision: to expand beyond the iPhone to third-party apps.
To make Siri available to additional apps -- so users, for instance, could check their bank balance or order a car service -- Apple needed to create a platform and coding tools that allows developers to integrate the virtual assistant into apps for the iPad, iPhone or Apple Watch. It did so in 2016.
Apple declined to make Mr. Stasior available for an interview. He didn't respond to emails seeking comment.
Siri co-founders Adam Cheyer and Dag Kittlaus left the team, citing personal reasons. They started a competing company, Viv Labs, to make a voice-based system available to third-party developers. Roughly a half dozen other members of the team followed. Samsung bought Viv last year for about $215 million.
In 2014, Apple moved Siri to a machine-learning system that used algorithms to improve its performance. Mr. Cue compared the new system to a brain transplant, saying it made Siri more versatile, allowing it to field requests no matter their phrasing. For example, it could answer both "Give me yesterday's Yankees score" and "What was yesterday's Yankees score?"
"The key about things like Siri is to make it not feel like something you have to think about," Mr. Cue said. Apple said Siri's error rate was cut in half.
Around the same time, Amazon introduced Alexa and its Echo speaker. The company pioneered a new category, blanketing television with quirky ads featuring Alec Baldwin. It sold an estimated 11 million home speakers in two years.
The device's popularity caught Apple off guard. It was designed to not only play music and answer questions but also control light switches and other home appliances, a concept Apple had recently introduced through a software system called HomeKit that allows developers to design thermostats and door locks that could be controlled by Siri.
The iPhone, in many ways a huge advantage, was also becoming an impediment. Apple says more than 375 million of its devices access Siri each month across 21 languages, and that Siri fields nearly 2 billion requests a week. In the U.S. alone, it has 70 million unique users weekly. By comparison, Amazon's Alexa is available only in German and English, and Google Assistant is available only in seven languages.
At the same time, the iPhone -- accounting for two thirds of sales -- so dominates internal focus that Apple largely abandoned the original Siri team's vision for an assistant that would go beyond calendar appointments and text messaging.
Instead, Apple added languages, added a male voice to Siri's original female voice, and allowed users to verbally create reminders while reading an article or viewing something on the iPhone. Apple and other large companies "tend to make improvements at the margin," said Larry Gillick, a former Siri speech scientist who serves as chief technology officer at a startup called Semantic Machines.
Amazon and Google, which introduced its own assistant in 2016, also enjoyed advantages over Siri because they had more data from their robust search engines to train their assistants and less-restrictive privacy policies than Apple, former Siri employees said.
Apple protects user privacy by randomly tagging Siri searches and keeping the data tagged for only six months, unlike Google and Amazon, which keep data until users ask for it to be discarded. The practice has complicated efforts to improve Siri because Apple relinquished control of data before it could be used to gauge the impact of software tweaks, former Siri engineers said.
"You're hamstrung," said Jason Douglas, a former member of the Siri search team. "The iTunes store has great data but the scale of it is not Google or Amazon."
Mr. Cue said Apple often uses generic data rather than user data to train its systems and has the ability to improve Siri's performance for individual users with information kept on their iPhones.
After Siri's brain transplant, plans to improve Siri's conversational ability stalled. Apple tasked Alex Acero, an expert in language processing, with combining two rival teams: speech recognition and natural language. The first involves adapting software to recognize spoken words, and the second involves interpreting those words. Following an executive-level power struggle, the project was transferred to another leader without expertise in the field, said Chuck Wooters, who was on the speech-recognition team.
The change disillusioned some speech-recognition experts. One left for Google and another for Amazon, said Mr. Wooters, who also departed and joined the Semantic startup, which is developing technology to make virtual assistants more conversational.
Apple declined to comment. Mr. Acero, who still works on Siri, didn't return requests for comment.
Other employees were discouraged by the reluctance to allow third-party apps to employ Siri. Apple added the ability to use Fandango to buy movie tickets in 2013, but Mr. Stasior tabled a plan to extend Siri's abilities to more apps, two former employees said. By the time the project moved forward two years later, Apple had reduced the number of new commands developers would be able to use, a former engineer said.
In June of last year, Apple unveiled about 150 new commands, called intents, for Siri at its annual developer's conference. It opened Siri to about seven types of apps, including payment and ride-sharing apps such as Venmo and Uber. By comparison, Amazon's Alexa allows developers to create custom commands, which it calls skills, and some 12,000 have been created, allowing users to order coffee, start a guided meditation or check their bank-account balance.
The limited scope of Siri's commands disappointed many developers, said Brian Roemmele, a developer who attended the announcement. "People went from being happy and excited to sitting in workshops and realizing, 'I can't use it,' " he said. "Some went back to that attitude: Siri's always going to be dumb. They moved on to developing for Google and Alexa."
Mr. Cue said Apple chose to open Siri to apps that people use frequently and make it versatile enough to understand requests said in a variety of ways. "These are things you do every day and use all the time," he said.
Apple on Monday announced developers would be able to use Siri for four more types of apps, including notes and bank accounts.
Siri's capabilities have lagged behind those of rivals elsewhere, as well. In tests across 5,000 different questions, it answered accurately 62% of the time, lagging the roughly 90% accuracy rate of Google Assistant and Amazon's Alexa, according to Stone Temple, a digital marketing firm.
A separate study by Loup Ventures, a market-research firm, shows Siri performs better than rivals on core iPhone functions, so-called command-related queries -- making calendar appointments, placing phone calls, sending text messages -- but doesn't do as well answering questions accurately from the web.
Apple has tried to close the gap through acquisitions. In 2015, it purchased VocalIQ, a Cambridge, England-based startup that designed a system to improve a virtual assistant's conversational ability. The plan was to feed Apple's data through the system to make Siri more conversational, a former engineer said, allowing a user to ask for a nearby coffee shop and subsequently narrow the list by asking for one with Wi-Fi.
Google Assistant has begun to offer those conversational capabilities, but Siri still hasn't made that advance. "I don't think anyone does an A+ on conversation," said Mr. Cue. "It's a challenging problem and there's a lot of work to be done in that area. It will get a lot better and needs to get a lot better."
Write to Tripp Mickle at Tripp.Mickle@wsj.com
(END) Dow Jones Newswires
June 07, 2017 11:39 ET (15:39 GMT)