Israeli startup Temi aims to put robots into the home

Yossi Wolf Photo; Eyal Izhar
Yossi Wolf Photo; Eyal Izhar

Yossi Wolf dreamed of a robot to help families communicate with elderly parents and managers to talk to employees, so he made one.

Near the edge of the old Tel Aviv bus terminal area stands a luxury tower, seemingly quite at odds with its surroundings, in which the offices of Temi, a startup, are located. Engineers are bending over a skeleton of a robot the height of a kitchen table or a child and in various stages of disassembly are in every corner. Wires connected to machines and huge screens, part of the brain behind the first Israeli home robot, are running lines of artificial intelligence code.

Behind one of the windows are placed previous versions of Temi in colors and a design that did not reach the production floor. Several meters away, we see the package for Temi - black and imposing, the size of a mini-refrigerator: complete, ready for operation, and designed for delivery to the first customers. The clock is ticking away the time to the date on which the company has promised to supply the robot - towards Christmas and the end of 2018. Shortly after that, at the huge CES exhibition in Las Vegas in early January, will come the official launching party.

The host, Temi founder and CEO Yossi Wolf, moves between the rooms and stands, proudly presenting the robot's internal organs. The name Temi is engraved again and again on the printed circuits winding their way between the sides. "Other than the cameras and the screen, almost all of the developments and designs are ours; only production is conducted in China," Wolf says. "600 robots have already been made, and 10,000 more are waiting to be made. We have open orders from distributors. We're already displaying them at Macy's in New York and the B8ta store chain all over the US."

The one-meter high, 12-kilogram Israeli robot is a combination of a tablet, a set of loudspeakers, and a charging surface on wheels. It is capable of following the users all over the house or office, and is equipped with Google Assistant software that turns it into a mobile assistant capable of following users anywhere in the house and speaking to them. It is like a version on wheels of smart speakers - products such as Google Home, Amazon Echo, and Portal, which Facebook launched this month, which have become very popular in recent years. The market price, however, is substantially higher than that of smart speakers: $1,500 per unit.

Wolf refuses to disclose how many robots have actually been ordered or sold. His expectations, however, are high: 10,000 units by the end of the year and 100,000 by the end of next year. The robot will be marketed initially only in the US, but within several months, sales will also be expanded to Asia, including China, while Temi is planning to reach Europe and Israel in the first half of 2019. "We're working all over the world right now, closing deals and agreements with distributors," Wolf says.

The robot, he explains, is designed to be multi-purpose. Temi envisions it helping elderly people to communicate with their families and buying products for them, keeping an eye on the kids, participating in team meetings in place of an absent employee, helping a doctor "visit" patients at home, and leading guests to a table in a restaurant. Wolf says that universities and hospitals are already interested in the product, as well as stores thinking of it as sales promotion.

He was focusing on military development until Grandma visited

Wolf, 38, founded Temi after a career in robotics that began in the air force and continued at Ehud Gal's ODF Optronics, in which he was VP robotics division. Later, in 2009, he joined forces with Elad Levy in founding military robots company Roboteam with their own money. "Venture capital funds make almost no investments in the hardware industry or person security," says Wolf. "The working assumption is that you won't be Elbit Systems 2, and if you look at the figures, there have been very few good defense exits - more troubles than big successes. We decided to do it nevertheless. We were crazy enough to jump into the fire and talented enough to escape it."

Roboteam now has five operational products sold in 30 countries. Its sales are estimated at $15 million a year. It works with the US air force, among others, and has raised over $60 million. Wolf, however, decided that it was time for a change in direction: "One day, I visited my grandmother. She was 90, and her hands started to tremble one day. She even had trouble drinking water, not to mention serving tea and cookies. So I told her, 'Grandma, I'm buying you a walker.' She was really angry, as if a product for old people was something awful for her, so I asked, 'Would you like a robot to help you?', and she said, "I want that.' That was the end of my market survey - I didn't need anyone to tell me that there are a lot of old people in the world."

While Levy went on leading Roboteam from the US, Wolf returned to Israel to found Temi in 2015. "I wanted to go from selling hundreds of robots in tenders to selling thousands of units, and from selling life-saving robots to selling life-changing robots," he says. "The change from military to civilian industry was a huge challenge. No robotics company broke into the civilian market and made a robot that really related to human beings. Robotics is mainly a profession for defense, factories, or the outer space industry."

Wolf nevertheless managed to convince investors that Temi had promising potential. He raised $58 million, and the most prominent investor was former Alibaba CTO John Wu. Temi now has 120 employees, of whom 100 are developers, some of whom came to Temi from Google, Waze, and Wix.

Actual manufacturing of the robot is by US company Flex, the world's third largest designer and manufacturer of electronics products. Wolf says that Flex "signed an agreement with Temi on strategy and gave it a line of credit, as it usually does with companies in which it believes." Production is taking place in a suburb of Shenzhen, China. Agreements were also signed with two big names for distributing the product: Fulfillment by Amazon and ModusLink.

What does the robot have against my jeans?

Simultaneously with progress in production and distribution, Wolf this year sent a prototype of the robot on a tour of major technology exhibitions. Among others, it was displayed at MWC, the world's largest mobile exhibition, where PCMag selected it as the "best robot" at the exhibition. It was later also displayed at the IFA exhibition in Berlin.

There is only one problem: the product, at least in the versions displayed to date, is having a tough time convincing people. Reviews in the professional magazines were unenthusiastic, and the demonstration held for "Globes" at Temi's offices in Tel Aviv was also rather disappointing. Temi managed to slowly follow Wolf, but had trouble following me. The CEO blamed my jeans, which were too dark for the robot to identify them - a weakness he said had already been fixed in a more advanced version than the one checked.

In a test we conducted, Temi's movement could be controlled by smartphone. It was able to conduct a video call, search for information on Google, and be asked to play music via voice commands. Temi contains a simple Android-based interface designed especially for it. According to the plan, many other apps will be added to the basic ones later. As of now, however, it offers fairly basic functions, and does not always understand what it hears on the first attempt (and only in English).

It is not as if Temi lacks advantages. Its design is original and quite elegant. It includes a tablet with cameras, a tray that also serves as a wireless charging surface, a set of high-quality Harman Kardon speakers, and all the navigational accessories that a robot needs: cameras, lasers, and a leader (the flashing light familiar from an autonomous vehicle), an original development by Temi. The problem is that with all due respect for Wolf's vision, in its current state, it is not really clear how the robot can be useful. A deadlier criticism than this asserts that it is merely a tablet on wheels, not something that justifies a $1,500 price tag.

Wolf is aware of the criticism. He suggests waiting for the launch of the final version, which he knows will also not be the final word. "We have a lot more hard work to do, and the idea is clear - constant improvement. Like a smartphone, a robot can be improved by upgrading the software," he says.

At the same time, he admits that it will be difficult to meet the standard expected by customers. "Behind the smartphone costing $1,000 that you hold in your hand are billions of dollars in investments. In order for Temi to achieve these standards, we have to create magic in the user experience and responsiveness. You are used to such high quality that if I fall below it, you say, 'It's not good enough.' How is it possible to compete with these billions with $5 million? We have to do a juggling act. As of now, we're creating and developing the robot with the capital we have, while at the same time raising more money for marketing and sales."

Promises, disappointments, and the next revolution

The problems encountered by Wolf are not confined to Temi. Home robots have been a favorite subject of books and sci-fi films for decades, but while the PC and industrial robots have improved, home robots have been left behind. Pepper, the robot made by Japanese company SoftBank Robotics, which is fairly similar to Temi in the way it functions, is still a niche product. Honda's Asimo mainly walks, and Sony's robot dog, Aibo, looks cute, but is quickly proving to be useless.

"There is an absurd situation in the market - a glass ceiling that has yet to be broken," Wolf says. "Google, Amazon, IBM - all of the big companies are dealing with it, but they haven't found a solution. Going from the kitchen to the living room is a very difficult task for robots, and opening a door is an almost unsolvable mission. It's like an autonomous car: the need and the advantages are clear, but the puzzle hasn't been solved yet. There's great technological complexity here, difficulties with the user experience, and commercialization difficulties. The market is lagging far behind what is the subject of university competitions and NASA."

Indeed, at the major technology exhibitions in the past two or three years, we repeatedly see robots that photograph well, but do little besides attracting attention, providing entertainment, and performing a limited number of basic communications actions. When the curtain falls, some of the projects are suspended or abandoned. In July 2018, Bosch closed down Mayfield Robotics, a startup that developed a home robot named Kuri. Honda called off its Asimo project a month before that.

Nevertheless, the big companies and startups are still trying to put robots in our home, carried aloft on the waves of three revolutions: three that have already occurred and one that is bound to happen. "The first revolution is the fall in hardware prices," Wolf says. "A camera that once cost $100,000 now costs $1. That's important, because a robot needs a lot of sensors, so the $1 million development cost for each robot has been reduced to hundreds of dollars. The second revolution is the increase in Internet speed, because without high-speed processing on the cloud, a lot of the algorithmics cannot be used. The third revolution is the smart assistants: Amazon Alaska, Microsoft Cortana, and Google Assistant. A robot is 1% hardware and 99% software."

According to Wolf, the industry is now waiting for 5G mobile communications networks, "which will make speeds 100 times as fast, thereby making it possible to perform a lot more algorithmics on the cloud and send the processing to the robot in real time. The connection speed is critical for robots; every half-second counts."

If Amazon is a competitor, it is good for Temi

There is something confusing in the classification of Temi as a robot. There is a good reason why Wolf is focusing on software, rather than hardware. There are no moving organs, and it is unable to make human gestures. The closest category seems to be a telepresence robot - robots for the business sector designed for mobile video conference calls. Like Temi, these robots following their operators and make it possible to conduct remote meetings in more flexible conditions than the expensive conference call devices in meeting rooms. There is even a store in Palo Alto in which the people selling video call robots are not present; they use the robots that they are selling to appear to the customers.

The price of robots with built-in screens of this type start at $2,100 and go as high as $15,000. Other robots close to Temi in structure are those made by Suzhou Pangolin Robot, which are designed like human dolls with a screen on their chest. Prices of these robots start at $3,000.

In the business sphere, in contrast to a comparison to standard home gadgets, Temi has a clear advantage in price. Wolf says that the cost of the robot component is less than $800, so that in the future, it will be possible to lower the price from the current $1,500. On the other hand, there are already cheap telepresence robots in the market that come with an anchoring device for a tablet instead of a screen.

Wolf prefers to describe Temi as an upgrade of a different non-robotic product category - smart loudspeakers based on voice assistants. He believes that he has something to offer here. The outstanding feature of Facebook Portal, for example, is that it follows what is taking place in a house through smart cameras. According to Wolf, this is a clumsy method that makes it necessary to know in advance to which camera to connect, while following through a mobile robot is simpler. "All of the major companies have entered this field; it's a world war," Wolf comments. "In our opinion, this is the future of the Internet. Every search will be based on voice assistants products. Fortunately, one of Amazon's projects in this sector is a home robot - Alaska is a robot scheduled to come out in late 2019. This is very good for us; instead of 'Why do I need this,' the question becomes 'What does it look like, and what will the price be?' It puts us in a much more attractive place from the perspective of companies like Google, Facebook, and Apple," he says, hinting at a desire to be acquired or enter strategic cooperation with one of the technology giants competing with Amazon.

Meanwhile, it appears that where the home consumer is concerned, Temi is concentrating mainly on early adopters - inveterate gadget freaks. The first people who bought smartphones sometimes paid thousands of shekels for slow, labored devices with bad screens - devices that could barely run two apps simultaneously and were outdated after only a few months. What did they get? Initial exposure to where the world was going.

The big question is when, if at all, the rest of the population will begin adopting robots like Temi and use them as babysitters for children, taking part in management meetings, and recommending courses at restaurants. With a $1,500 price tag, it is possible that robots that are mechanically simpler, quasi-advanced personal assistants will be available in the relatively near future. Less optimistically, all of the signs indicate that robots like the ones in science fiction will have to wait for many years.

Published by Globes, Israel business news - - on October 16, 2018

© Copyright of Globes Publisher Itonut (1983) Ltd. 2018

Yossi Wolf Photo; Eyal Izhar
Yossi Wolf Photo; Eyal Izhar
Twitter Facebook Linkedin RSS Newsletters גלובס Israel Business Conference 2018