All posts in “economy”

Andrew Ng is raising a $150M AI Fund


We knew that Andrew Ng had more than just a series of deep learning courses up his sleeve when he announced the first phase of his deeplearning.ai last week. It’s clear now that the turn of Ng’s three part act is a $150 million venture capital fund, first noted by PEHub, targeting AI investments.

Ng, who formerly founded Google’s Brain Team and served as chief scientist at Baidu has long evangelized the benefits AI could bring to the world. During an earlier conversation, Ng told me that his personal goal is to help bring about an AI-powered society. It would follow that education via his deep learning classes is one step of that and providing capital and other resources is another.

2017 has been a particularly active year for starting AI-focused venture capital funds. In the last few months we have seen Google roll out Gradient Ventures, Basis Set Ventures hall in $136 million, Element.AI raise $102 million, Microsoft Ventures start its own AI fund and Toyota corral $100 million for AI investment.

It’s unclear at this point how Ng’s AI Fund will differentiate from the pack. Many of these funds are putting time and resources into securing data sets, technical mentors and advanced simulation tools to support the unique needs of AI startups. Of course Ng’s name recognition and network should help ensure solid deal flow and enable Ng to poach and train talent for startups in need of scarce deep learning engineers.

I’ve sent a note to Andrew and we will update this post if and when we get more details.

Featured Image: Dawn Endico/Flickr UNDER A CC BY-SA 2.0 LICENSE

Machine learning can tell if you’re wearing swap-meet Louie


A wise man once said “The hat mighta had a L V on the back but at the swap meet that ain’t jack,” and now researchers can ensure that the Louis Vuitton or Prada or Coach you bought is the real deal. The system, which essentially learns the difference between real and fake products over time, uses a small microscope connected to a phone.

“The underlying principle of our system stems from the idea that microscopic characteristics in a genuine product or a class of products-corresponding to the same larger product line-exhibit inherent similarities that can be used to distinguish these products from their corresponding counterfeit versions,” said New York University Professor Lakshminarayanan Subramanian.

The researchers have commercialized the product as Entrupy Inc., a startup founded by Ashlesh Sharma, an NYU doctoral graduate, Vidyuth Srinivasan and Professor Subramanian. You can even buy the product now and run a few dozen authentications per month.

The system is non-invasive and does not damage the merchandise. Because it uses a “dataset of three million images” you can assess a material almost instantly. It takes about 15 seconds to test a product and it can distinguish fabrics, leather, pills, shoes and toys. It can even tell if electronics are authentic.

“The classification accuracy is more than 98 percent, and we show how our system works with a cellphone to verify the authenticity of everyday objects,” said Subramanian.

Entrupy has raised $2.6 million in funding and has apparently authenticated $14 million in real and fake purses, watches and other fancy stuff. I can definitely help out if you get angry and feel the need to begin sockin’ more fools than Patrick Swayze because they are selling bootleg purses.

[embedded content]

Gas pump card skimmer now phones home


In an unsurprising move by credit card thieves, police have found a new credit card skimmer that sends stolen data via SMS. By tearing apart cheap phones, crooks are able to send credit card information to their location instantly without having to access the skimmer physically or rely on an open Bluetooth connection.

Brian Krebs received images of the skimmer from an unnamed source. They were found at a gas station in the Northeast.

This skimmer connected to the internals of the pump and received power from the pump itself, meaning there was no need to worry about the battery failing. It’s unclear how this model worked but most likely it intercepts the credit card data as it’s being swiped. There’s still hope, however, because gas stations are trying to fight back.

“Many filling stations are upgrading their pumps to include more physical security — such as custom locks and security cameras. In addition, newer pumps can accommodate more secure chip-based payment cards that are already in use by all other G20 nations,” wrote Krebs.

He also recommends not using debit cards at all at gas stations and, obviously, protecting your PIN from prying eyes or cameras.

Clara Labs nabs $7M Series A as it positions its AI assistant to meet the needs of enterprise teams


Clara Labs, creator of the Clara AI assistant, is announcing a $7 million Series A this morning led by Basis Set Ventures. Slack Fund also joined in the round, alongside existing investors Sequoia and First Round. The startup will be looking to further differentiate within the crowded field of email-centric personal assistants by building in features and integrations to address the needs of enterprise teams.

Founded in 2014, Clara Labs has spent much of the last three years trying to fix email. When CC-ed on emails, the Clara assistant can automatically schedule meetings — reasoning around preferences like location and time.

If this sounds familiar, it’s because you’ve probably come across x.ai or Fin. But while all three startups look similar on paper, each has its own distinct ideology. Where Clara is running toward the needs of teams, Fin embraces the personal pains of travel planning and shopping. Meanwhile, x.ai opts for maximum automation and lower pricing.

That last point around automation needs some extra context. Clara Labs prides itself in its implementation of a learning strategy called human-in-the-loop. For machines to analyze emails, they have to make a lot of decisions — is that date when you want to grab coffee, or is it the start of your vacation when you’ll be unable to meet?

In the open world of natural language, incremental machine learning advances only get you so far. So instead, companies like Clara convert uncertainty into simple questions that can be sent to humans on demand (think proprietary version of Amazon Mechanical Turk). The approach has become a tech trope with the rise of all things AI, but Maran Nelson, CEO of Clara Labs, is adamant that there’s still a meaningful way to implement agile AI.

The trick is ensuring that a feedback mechanism exists for these questions to serve as training materials for uncertain machine learning models. Three years later, Clara Labs is confident that its approach is working.

Bankrolling the human in human-in-the-loop does cost everyone more, but people are willing to pay for performance. After all, even a nosebleed-inducing $399 per month top-tier plan costs a fraction of a real human assistant.

Anyone who has ever experimented with adding new email tools into old workflows understands that Gmail and Outlook have tapped into the dark masochistic part of our brain that remains addicted to inefficiency. It’s tough to switch and the default of trying tools like Clara is often a slow return to the broken way of doing things. Nelson says she’s keeping a keen eye on user engagement and numbers are healthy for now — there’s undoubtedly a connection between accuracy and engagement.

As Clara positions its services around the enterprise, it will need to take into account professional sales and recruiting workflows. Integrations with core systems like Slack, CRMs and job applicant tracking systems will help Clara keep engagement numbers high while feeding machine learning models new edge cases to improve the quality of the entire product.

“Scheduling is different if you’re a sales person and your sales team is measured by the total number of meetings scheduled,” Nelson told me in an interview.

Nelson is planning to make new hires in marketing and sales to push the Clara team beyond its current R&D comfort zone. Meanwhile the technical team will continue to add new features and integrations, like conference room booking, that increase the value-add of the Clara assistant.

Xuezhao Lan of Basis Set Ventures will be joining the Clara Labs board of directors as the company moves into its next phase of growth. Lan will bring both knowledge of machine learning and strategy to the board. Today’s Clara deal is one of the first public deals to involve the recently formed $136 million AI-focused Basis Set fund.

Reach Robotics closes $7.5M Series A for its augmented reality bots


After years of research and development, Reach Robotics has closed a $7.5 million Series A, co-led by Korea Investment Partners (KiP) and IGlobe, to bring its augmented reality bots to market in a big way. The Bristol-based startup is looking to expand into the U.S., and the team is exploring opportunities for growth into other European and Asian markets.

Reach Robotics’ first product, MekaMon, launched last fall. Today’s round comes after the company produced and sold an initial run of 500 of its four-legged, crab-like, bots. MekaMon fits into an emerging category of smartphone-enabled augmented reality toys like Anki.

Silas Adekunle, CEO of Reach Robotics, tells me the influx of capital will be used to make some strategic hires and increase brand recognition through marketing. This is the first time the startup has announced a funding round. Adekunle tells me his experience raising capital wasn’t easy; as they say, hardware is hard.

“It was hard to pitch in our early days because people didn’t believe,” explained Adekunle.

MekaMon sits somewhere between toy and full-fledged robot. Unlike the radio-controlled RadioShack robots of yesteryear, MekaMon costs a hefty $329. At first glance this can be hard to swallow, but Adekunle remains adamant that he is building a platform and not a line of toys — think PS4 instead of an expensive, single-use robot collecting dust on a shelf.

Outside of retail sales, another avenue for the company to make money is through partnerships within the entertainment industry. Adekunle says that Reach would never go out of its way to deliver a specific product for a client, but he always keeps an eye out for overlap where a partnership could occur with minimal operational changes.

“People are taken aback that something could be this realistic,” asserts Adekunle. “If you strip back the product and lose that, then you don’t have an innovative company.”

Because Reach is selling software-enabled hardware, it has the opportunity to collect all sorts of interesting data that it can use to fine-tune its products. The startup is able to track retention in aggregate and look at how people actually use their robots. Moreover, if MekaMon suffers leg failure, Reach can analyze indicators like temperature readings and torque.

Adekunle insists on keeping the Reach Robotics team interdisciplinary — one employee helped shape the way robots move in the Transformers movie series. This same team is focused on empowering the next group of developers who will build on the MekaMon platform and create new use cases, beyond the company’s initial vision for the product.