Labelbox raises $10 million for its services to support machine learning applications

Published April 9, 2019

Labelbox, a provider of services to create, manage and maintain data sets for machine learning applications, has raised $10 million in a new round of funding.

The financing came from Gradient Ventures, Google’s AI-focused venture fund, with participation from previous investors Kleiner Perkins, First Round Capital and Sumon Sadhu, an angel investor.

Labelbox manages the process of outsourcing data labeling for organizations and provides toolkits for companies or organizations to manage the data they’re receiving and ensuring the quality of that data, according to chief executive Manu Sharma.

For the Labelbox founders — Sharma; Dan Rasmuson, the company’s chief technology officer; and Brian Rieger, the chief operating officer — the tools they developed are simply an extension of the services they’d needed at their previous employers — companies like DroneDeploy, Planet Labs and Boeing.

Financing from the round will be used to double the size of its team from 11 employees to 22, and build out its sales and marketing teams.

Labelbox counts around 50 customers for its service and charges them based on the volume of data that companies upload and the breadth of services they use, Sharma said. Some named customers include FLIR Systems, Lytx, Airbus, Genius Sports and KeepTruckin.

As we’d reported when Labelbox launched from stealth last year, anyone can use the company’s toolkit for free. Companies are charged once they hit a certain usage threshold. Lytx, for instance, uses Labelbox for its DriveCam, a system installed on half a million trucks with cameras that use AI to detect unsafe driver behavior so they can be coached to improve. And the media and publishing giant Conde Nast is using Labelbox to match runway fashion to related items in their archive of content.

“Labelbox substantially reduces model development times and empowers data science teams to build great machine learning applications,” said Sharma in a statement. “With the new funding, Labelbox will continue to double down on bringing data labeling infrastructure to the machine learning teams with powerful automation, collaboration, and enterprise-grade features.”

Gradient Ventures was interested enough in the technology to invest, and sees promise in the company’s ability to support the development of machine learning tools globally.

“Labelbox is well-positioned to fuel the industrialization of machine learning across many sectors, such as manufacturing, transportation and healthcare. In doing so, they will unlock the potential of AI for companies across the globe,” said Anna Patterson, founder and managing partner at Gradient Ventures.

Sapling, an employee management and on-boarding platform, lands $4 million in seed funding

Published February 26, 2019

Sapling, a three-year-old, San Francisco-based company whose employee management and onboarding software is being adopted by a small but growing number of mid-size companies with far-flung workforces, is announcing today that it has raised $4 million in funding from Gradient Ventures, which is Google’s AI fund, and Tuesday Capital, formerly known as CrunchFund.

It quietly secured the funding several months ago and has been using it to ramp up to the 50 people it currently employs.

The company’s founding team is the kind that investors like to see, meaning that in many ways, their previous work experiences led them to start Sapling .

Co-founder and CEO Bart Macdonald has spent his entire career in HR, working most recently in Melbourne, Australia, as a regional director for the global coding school General Assembly, where he hired and managed a 10-person marketing, sales and operations team.

Meanwhile, co-founder Andy Crebar (born in the same Sydney hospital as Macdonald, a day later) also knows the plight of individuals trying to seamlessly onboard new hires, having worked most recently on business development initiatives at a fintech startup called Credible Labs, where adding headcount was, as at many companies, a point of frustration.

“I liked that Bart and Andy had lived through their own experiences dealing with crappy HR software in previous positions and thus really understood how customers view the problem,” says Tuesday Capital co-founder Pat Gallagher. “The fact that neither are technical would have been an issue if we were investing pre product, but by the time we invested, they had proven they could build software that their customers loved.”

In fact, says Gallagher, his team was drawn to Sapling specifically because a handful of the firm’s portfolio companies has been using its onboarding software and “really raving about it. It’s hard to find HR software that people really like, so that was a big positive for us and helped cut through the noise of the space that they operate in.”

So what’s so special about Sapling? Mostly, it seems, its approach brings together the tools and software that HR execs are already using, including ADP for payroll, or G Suite for productivity, and Lever for recruiting, integrations that also employ a heavy dose of AI to anticipate the behaviors of employees, making it easier for managers to recruit, aid, manage and support current and future staffers.

As Macdonald explains it on the simplest level, Sapling not only provisions software for them but it connects their tools “so they don’t have to open 10 tabs. All they have to do is run their workflow inside of Sapling so that, for example, an employee can ask for time off in Slack,” and that request will automatically be reflected in the employer’s payroll and benefits systems (once approved).

Sapling currently works with companies with anywhere from 100 to 1,500 employees, including InVision, an eight-year-old commercial platform used by design teams to create digital products for mobile and desktop that is currently investing its Series F round. InVision, which has a large distributed workforce, says Sapling has saved the company 1,000 hours by speeding up communications and making employee engagement far more seamless.

What comes next for Sapling remains to be seen. It’s in an awfully crowded category, with no shortage of all-in-one HR solutions attracting venture capital. In the meantime, with low unemployment creating headaches for many outfits looking to keep its talent, Sapling is smartly positioning itself as an important tool in specifically helping companies with geographically distributed teams to retain and engage employees. Customers like InVision, along with Digital Ocean, KPMG and Kayak, say it’s working, too.

Above, left to right: founders Bart Macdonald and Andy Crebar, courtesy of Sapling.

Gradient Ventures, Google’s AI fund, leads $7M investment in English learning app Elsa

Published February 26, 2019

Google’s Gradient Ventures, the search giant’s dedicated AI fund, is casting its eye to Asia after it led a $7 million Series A round for Elsa, a startup that operates an app for English language learners.

The deal is Gradient’s first in Asia, and it includes participation from existing investors Monk’s Hill Ventures and SOSV. Elsa has now raised $12 million to date.

Elsa was founded in 2015 as a way to help non-English speakers improve their accent and general speaking ability. Vu Van, CEO and one half of the founding team, is a Vietnamese national who, despite being fluent in English, struggled to be understood after moving to the U.S. to study and then work. Together with speech recognition researcher Dr. Xavier Anguera — the startup’s CTO who leads its Portugal-based tech team — Van started Elsa to help people in the same predicament.

“I was very good at grammar, reading and writing but I realized people had a hard time understanding me because I had a very strong accent and my pronunciation wasn’t proper,” Van, who is based in San Francisco but travels extensively, told TechCrunch in an interview. “This impacts confidence when you apply for jobs or are even just meeting friends.”

“There are so many English learning solutions but they are mostly focused on expanding vocabulary or grammar, very few deal with pronunciation,” she added.

Elsa uses voice recognition and AI to grade a user’s speaking versus standard American English (and I thought us Brits were the global standard…) giving them a score at the end. That helps track their progress, while it focuses on pronunciation with a detailed review on how a user is speaking.

The service uses a freemium model that grants users full access to 1,000 courses for around $3-6 per month depending on the length of the package they select. That ranges from one month of access to 12 months. New content is added every week, Van said.

With this money in the bag, Elsa is going after growth in a number of its most promising markets.

The service has users in more than 100 countries, but Vietnam is its top market, with two million paying users. Partly because it is Van’s home market, Elsa has doubled down on Vietnam with a local sales team and localized payments, including the likes of bank transfers and local wallets.

That’s the blueprint for expansion in its next three target countries: Japan, Indonesia and India. Elsa has opened an office in Tokyo and is planning to introduce more localized content for Japanese users. Similar efforts will happen in Indonesia and India, where Van said the app sees strong engagement and downloads without any paid marketing efforts.

Elsa is also working on expanding its content from English to include other languages. Spanish is currently on the horizon and the company is already preparing the back-end technology to make it possible.

“We have to build the voice recognition technology to recognize those languages accurately. We have the infrastructure but now just need to collect voice data to train the model,” explained Van.

Vu Van started Elsa in 2015 with Dr. Xavier Anguera to help non-English speakers improve their accent and general speaking ability.

Beyond geographic expansion, Elsa is also going after schools and classrooms. Already, in Vietnam, it is working with a handful of schools that have added the app to their classroom work. The company allows schools to upload their specific content or curriculum to Elsa to make it part of a student’s homework or assessment. Teachers can see if a student has completed oral homework, and the app grades their efforts.

“We want to help these teachers help their students,” Van said. “Even with the best intentions, they simply can’t teach speaking.”

The model for the education push sees schools pay a licensing fee per student, which Van said is subsidized, while uploading their content is free.

Snagging investment from Gradient is a notable achievement for Elsa, but it will also allow the startup to tap into the company’s talent, too. That’s because Gradient operates a rotational program that allows Google employees to spend three to six months working at portfolio startups on secondment. That process hasn’t kicked off for Elsa just yet, but Van is hopeful of securing an engineer who might otherwise be prohibitively expensive for her company.

Gradient Ventures was founded in 2017 and this deal is the fund’s 18th, according to Crunchbase. Its previous investments include Canvass Analytics and Test.ai.

The Elsa team