CBPO

Tag: Builds

Facebook has acquired Servicefriend, which builds ‘hybrid’ chatbots, for Calibra customer service

September 22, 2019 No Comments

As Facebook prepares to launch its new cryptocurrency Libra in 2020, it’s putting the pieces in place to help it run. In one of the latest developments, it has acquired Servicefriend, a startup that built bots — chat clients for messaging apps based on artificial intelligence — to help customer service teams, TechCrunch has confirmed.

The news was first reported in Israel, where Servicefriend is based, after one of its investors, Roberto Singler, alerted local publication The Marker about the deal. We reached out to Ido Arad, one of the co-founders of the company, who referred our questions to a team at Facebook. Facebook then confirmed the acquisition with an Apple-like non-specific statement:

“We acquire smaller tech companies from time to time. We don’t always discuss our plans,” a Facebook spokesperson said.

Several people, including Arad, his co-founder Shahar Ben Ami, and at least one other indicate that they now work at Facebook within the Calibra digital wallet group on their LinkedIn profiles. Their jobs at the social network started this month, meaning this acquisition closed in recent weeks. (Several others indicate that they are still at Servicefriend, meaning they too may have likely made the move as well.)

Although Facebook isn’t specifying what they will be working on, the most obvious area will be in building a bot — or more likely, a network of bots — for the customer service layer for the Calibra digital wallet that Facebook is developing.

Facebook’s plan is to build a range of financial services for people to use Calibra to pay out and receive Libra — for example, to send money to contacts, pay bills, top up their phones, buy things and more.

It remains to be seen just how much people will trust Facebook as a provider of all these. So that is where having “human” and accessible customer service experience will be essential.

“We are here for you,” Calibra notes on its welcome page, where it promises 24-7 support in WhatsApp and Messenger for its users.

Screenshot 2019 09 21 at 23.25.18

Servicefriend has worked on Facebook’s platform in the past: specifically it built “hybrid” bots for Messenger for companies to use to complement teams of humans, to better scale their services on messaging platforms. In one Messenger bot that Servicefriend built for Globe Telecom in the Philippines, it noted that the hybrid bot was able to bring the “agent hours” down to under 20 hours for each 1,000 customer interactions.

Bots have been a relatively problematic area for Facebook. The company launched a personal assistant called M in 2015, and then bots that let users talk to businesses in 2016 on Messenger, with quite some fanfare, although the reality was that nothing really worked as well as promised, and in some cases worked significantly worse than whatever services they aimed to replace.

While AI-based assistants such as Alexa have become synonymous with how a computer can carry on a conversation and provide information to humans, the consensus around bots these days is that the most workable way forward is to build services that complement, rather than completely replace, teams.

For Facebook, getting its customer service on Calibra right can help it build and expand its credibility (note: another area where Servicefriend has build services is in using customer service as a marketing channel). Getting it wrong could mean issues not just with customers, but with partners and possibly regulators.


Social – TechCrunch


Facebook has acquired Servicefriend, which builds ‘hybrid’ chatbots, for Calibra customer service

September 22, 2019 No Comments

As Facebook prepares to launch its new cryptocurrency Libra in 2020, it’s putting the pieces in place to help it run. In one of the latest developments, it has acquired Servicefriend, a startup that built bots — chat clients for messaging apps based on artificial intelligence — to help customer service teams, TechCrunch has confirmed.

The news was first reported in Israel, where Servicefriend is based, after one of its investors, Roberto Singler, alerted local publication The Marker about the deal. We reached out to Ido Arad, one of the co-founders of the company, who referred our questions to a team at Facebook. Facebook then confirmed the acquisition with an Apple-like non-specific statement:

“We acquire smaller tech companies from time to time. We don’t always discuss our plans,” a Facebook spokesperson said.

Several people, including Arad, his co-founder Shahar Ben Ami, and at least one other indicate that they now work at Facebook within the Calibra digital wallet group on their LinkedIn profiles. Their jobs at the social network started this month, meaning this acquisition closed in recent weeks. (Several others indicate that they are still at Servicefriend, meaning they too may have likely made the move as well.)

Although Facebook isn’t specifying what they will be working on, the most obvious area will be in building a bot — or more likely, a network of bots — for the customer service layer for the Calibra digital wallet that Facebook is developing.

Facebook’s plan is to build a range of financial services for people to use Calibra to pay out and receive Libra — for example, to send money to contacts, pay bills, top up their phones, buy things and more.

It remains to be seen just how much people will trust Facebook as a provider of all these. So that is where having “human” and accessible customer service experience will be essential.

“We are here for you,” Calibra notes on its welcome page, where it promises 24-7 support in WhatsApp and Messenger for its users.

Screenshot 2019 09 21 at 23.25.18

Servicefriend has worked on Facebook’s platform in the past: specifically it built “hybrid” bots for Messenger for companies to use to complement teams of humans, to better scale their services on messaging platforms. In one Messenger bot that Servicefriend built for Globe Telecom in the Philippines, it noted that the hybrid bot was able to bring the “agent hours” down to under 20 hours for each 1,000 customer interactions.

Bots have been a relatively problematic area for Facebook. The company launched a personal assistant called M in 2015, and then bots that let users talk to businesses in 2016 on Messenger, with quite some fanfare, although the reality was that nothing really worked as well as promised, and in some cases worked significantly worse than whatever services they aimed to replace.

While AI-based assistants such as Alexa have become synonymous with how a computer can carry on a conversation and provide information to humans, the consensus around bots these days is that the most workable way forward is to build services that complement, rather than completely replace, teams.

For Facebook, getting its customer service on Calibra right can help it build and expand its credibility (note: another area where Servicefriend has build services is in using customer service as a marketing channel). Getting it wrong could mean issues not just with customers, but with partners and possibly regulators.


Startups – TechCrunch


Dynamic Yield, which builds Amazon-like personalisation for the rest of us, raises $38M

November 3, 2018 No Comments

Amazon, one of the world’s largest companies, has transformed the face of commerce in part because it has managed at once to be “The Everything Store” but still with a route into its sea of products that, for most users, surfaces what they might most want to see (and importantly buy or consume). That kind of personalisation has become a goal not just for e-commerce companies, but for any organization running a digital business: users are constantly distracted, and when their attention is caught, they do not want to spend time figuring out what they most want.

Not every business is Amazon, though, so we are seeing a crop of startups emerging that are working on ways to help the rest of the digital world be just as optimised and personalised as Amazon. Now one of them, an Israeli startup called Dynamic Yield, has raised more money as it continues to expand its business, both to more platforms and to more geographies.

The startup’s Series D has now closed off at $ 38 million, with the inclusion of a $ 5 million strategic investment from Naver, Korea’s “Google” (it’s the country’s top search portal) that is also behind messaging apps Line and Snow. The plan is for Naver to help bring Dynamic Yield to Korea and Japan, by incorporating its tech into its own services and those of others that work with Naver.

(Personalisation and aggregators are strong magnets for users in Asia and thus big magnets for funding: ByteDance, which provides news aggregation among other services, was recently valued at $ 75 billion.)

Naver is not the only search engine that has caught sight of Dynamic Yield over the years. Previous investors include Baidu (“the Google of China”), and we’ve heard that when the startup was younger — it was founded in 2011 — Google had tried to acquire it (Dynamic Yield rejected the offer, and it’s been approached for acquisitions numerous times since then).

Other strategic investors include The New York Times and Deutsche Telekom, alongside other backers like Innovation Endeavors, Bessemer Venture Partners, Marker Capital and more.

Dynamic Yield has raised $ 85 million to date and is now valued at “hundreds of millions of dollars,” but less than $ 500 million, a source at the company said, after seeing a strong expansion of its services. 

Dynamic Yield says it works with more than 220 global brands, and its tech reaches 600 million unique users each month, across 10 billion page views and 600 billion “events” on those pages. It claims its AI-based personalisation technology can lift revenues (or other engagement metrics) by 10-15 percent. 

“It makes us an effective tool for surviving in a market where customer acquisition cost keeps getting more expensive,” co-founder and CEO Liad Agmon said in an interview.

Dynamic Yield doesn’t talk about many of its customers on the record — most don’t like to reveal to rivals who they work with, Agmon said.

But they include a number of big brands across e-commerce, travel, finance, media and other segments that use its tech not just to show more targeted products to prospective shoppers, but to help power advertising, recommend content and position the same information to different people in different ways depending on who is viewing it (for example with different headlines).

There are a lot of personalisation and A/B analytics companies in the market today — others include Adobe, Marketo (which is becoming a part of Adobe), Optimizely and many more. Indeed, I’d be very surprised if Amazon is not working on ways of productising its own personalisation tech in a way that is not intrinsically linked to its own marketplace (because some will never want to sell there, and because personalisation can be used for so much more than just e-commerce).

Dynamic Yield, however, claims that it has an edge over these because of how it works.

Agmon says that the tech sits on top of whichever CMS or other backend server that a site is using and is activated by way of a small amount of code. It uses machine learning to both “read” what is in a site, and matches that up against specific visitors and its own trove of experience.

Agmon added that when a business already has information about that visitor, that is the primary data that is used; otherwise it also incorporates other data sources like Acxiom and others — much the way that other marketing tech does — to form a stronger picture of your tastes.

It then runs this data through its own machine learning algorithms both to recommend content and to help a marketing manager figure out better customer segmentation overall. There is an “autopilot” version of the product where everything is automated based on Dynamic Yield’s algorithms; or options to use the data sources to set up specific marketing campaigns; or (as is common) a combination of the two.

Going forward, Agmon said the plan is to work across an increasing number of interfaces where customers are going today to discover and buy goods and services. Indeed, we’ve described how some of the newest e-commerce startups have eschewed any website or app of their own and work exclusively in third-party messaging apps to acquire customers and sell goods.

But it’s not just these new digital platforms that are becoming targets for personalisation startups like Dynamic Yield.

Agmon said that his company is also working with a major retailer that is using its tech at its in-person payment points. When — for example — a customer comes to order a latte, instead of generic upselling to the latest seasonal flavour, the person taking the order will now know if the customer ever orders a sweet injection, or if she/he is more of a savoury snack sort of person. The cashier will then know what to recommend to eat with that drink that is more likely to be purchased.

The mom-and-pop shop with its reputation for knowing the regulars and what they like might have found its dystopian (but useful) heir.


Enterprise – TechCrunch


Talla builds a smarter customer knowledge base

August 21, 2018 No Comments

Talla is taking aim at the customer service industry with its latest release, an AI-infused knowledge base. Today, the company released version 2.0 of the Talla Intelligent Knowledge Base.

The company also announced that Paula Long, most recently CEO at Data Gravity, has joined the company as SVP of engineering.

This tool combines customer content with automation, chatbots and machine learning. It’s designed to help teams who work directly with customers get at the information they need faster and the machine learning element should allow it to improve over time.

You can deploy the product as a widget on your website to give customers direct access to the information, but Rob May, company founder and CEO says the most common use case involves helping sales, customer service and customer success teams get access to the most relevant and current information, whether that’s maintenance or pricing.

The information can get into the knowledge base in several ways. First of all you can enter elements like product pages and FAQs directly in the Talla product as with any knowledge base. Secondly if an employee asks a questions and there isn’t an adequate answer, it exposes the gaps in information.

Talla Knowledge Base gap list. Screenshot: Talla

“It really shows you the unknown unknowns in your business. What are the questions people are asking that you didn’t realize you don’t have content for or you don’t have answers for. And so that allows you to write new content and better content,” May explained.

Finally, the company can import information into the knowledge base from Salesforce, ServiceNow, Jira or wherever it happens to live, and that can be added to a new page or incorporated into existing page as appropriate.

Employees interact with the system by asking a bot questions and it supplies the answers if one exists. It works with Slack, Microsoft Teams or Talla Chat.

Talla bot in action in Talla Chat. Screenshot: Talla

Customer service remains a major pain point for many companies. It is the direct link to customers when they are having issues. A single bad experience can taint a person’s view of a brand, and chances are when a customer is unhappy they let their friends know on social media, making an isolated incident much bigger. Having quicker access to more accurate information could help limit negative experiences.

Today’s announcement builds on an earlier version of the product that took aim at IT help desks. Talla found customers kept asking for a solution that provided similar functionality with customer-facing information and they have tuned it for that.

May launched Talla in 2015 after selling his former startup Backupify to Datto in 2014. The company, which is based near Boston, has raised $ 12.3 million.


Enterprise – TechCrunch


Leena AI builds HR chatbots to answer policy questions automatically

June 30, 2018 No Comments

Say you have a job with a large company and you want to know how much vacation time you have left, or how to add your new baby to your healthcare. This usually involves emailing or calling HR and waiting for an answer, or it could even involve crossing multiple systems to get what you need.

Leena AI, a member of the Y Combinator Summer 2018 class, wants to change that by building HR bots to answer questions for employees instantly.

The bots can be integrated into Slack or Workplace by Facebook and they are built and trained using information in policy documents and by pulling data from various back-end systems like Oracle and SAP.

Adit Jain, co-founder at Leena AI, says the company has its roots in another startup called Chatteron, which the founders started after they got out of college in India in 2015. That product helped people build their own chatbots. Jain says along the way, they discovered while doing their market research a particularly strong need in HR. They started Leena AI last year to address that specific requirement.

Jain says when building bots, the team learned through its experience with Chatteron that it’s better to concentrate on a single subject because the underlying machine learning model gets better the more it’s used. “Once you create a bot, for it to really add value and be [extremely] accurate, and for it to really go deep, it takes a lot of time and effort and that can only happen through verticalization,” Jain explained.

Photo: Leena AI

What’s more, as the founders have become more knowledgeable about the needs of HR, they have learned that 80 percent of the questions cover similar topics, like vacation, sick time and expense reporting. They have also seen companies using similar back-end systems, so they can now build standard integrators for common applications like SAP, Oracle and NetSuite.

Of course, even though people may ask similar questions, the company may have unique terminology or people may ask the question in an unusual way. Jain says that’s where the natural language processing (NLP) comes in. The system can learn these variations over time as they build a larger database of possible queries.

The company just launched in 2017 and already has a dozen paying customers. They hope to double that number in just 60 days. Jain believes being part of Y Combinator should help in that regard. The partners are helping the team refine its pitch and making introductions to companies that could make use of this tool.

Their ultimate goal is nothing less than to be ubiquitous, to help bridge multiple legacy systems to provide answers seamlessly for employees to all their questions. If they can achieve that, they should be a successful company.


Enterprise – TechCrunch


DeepCoder builds programs using code it finds lying around

February 23, 2017 No Comments

markus-spiske-207946 Like all great programmers I get most of my code from StackOverflow questions. Can’t figure out how to add authentication to Flask? Easy. Want to shut down sendmail? Boom. Now, thanks to all the code on the Internet, a robot can be as smart as a $ 180,000 coder. The system, called DeepCoder, basically searches a corpus of code to build a project that works to spec. It’s been used… Read More
Enterprise – TechCrunch


Man Builds A Real, Working Wall-E That’s Still Eternally Hunting For Eve

January 1, 2013 No Comments

Biggs is the East Coast Editor of TechCrunch. Biggs has written for the New York Times, InSync, USA Weekend, Popular Mechanics, Popular Science, Money and a number of other outlets on technology and wristwatches. He is the former editor-in-chief of Gizmodo.com and lives in Bay Ridge, Brooklyn. You can Tweet him here and G+ him here. Email him directly at… → Learn More

Mike Senna, a California-based roboticist, has built a real, working Wall-E that can move around, wave, and call out his own cute name in a rattly, digitized voice. Mike is the guy who built a real, working R2-D2 and his latest project is a real masterpiece of animatronics and robotic motion.

Read More