Committed to privacy, Snips founder wants to take on Alexa and Google, with blockchain

Earlier this year we saw the headlines of how the users of popular voice assistants like Alexa and Siri and continue to face issues when their private data is compromised, or even sent to random people. In May it was reported that Amazon’s Alexa recorded a private conversation and sent it to a random contact. Amazon insists its Echo devices aren’t always recording, but it did confirm the audio was sent.

The story could be a harbinger of things to come when voice becomes more and more ubiquitous. After all, Amazon announced the launch of Alexa for Hospitality, its Alexa system for hotels, in June. News stories like this simply reinforce the idea that voice control is seeping into our daily lives.

The French startup Snips thinks it might have an answer to the issue of security and data privacy. Its built its software to run 100% on-device, independently from the cloud. As a result, user data is processed on the device itself, acting as a potentially stronger guarantor of privacy. Unlike centralized assistants like Alexa and Google, Snips knows nothing about its users.

Its approach is convincing investors. To date, Snips has raised €22 million in funding from investors like Korelya Capital, MAIF Avenir, BPI France and Eniac Ventures. Created in 2013 by 3 PhDs, and now employing more than 60 people in Paris and New York, Snips offers its voice assistant technology as a white-labelled solution for enterprise device manufacturers.

It’s tested its theories about voice by releasing the result of a consumer poll. The survey of 410 people found that 66% of respondents said they would be apprehensive of using a voice assistant in a hotel room, because of concerns over privacy, 90% said they would like to control the ways corporations use their data, even if it meant sacrificing convenience.

“Сonsumers are increasingly aware of the privacy concerns with voice assistants that rely on cloud storage — and that these concerns will actually impact their usage,” says Dr Rand Hindi, co-founder and CEO at Snips. “However, emerging technologies like blockchain are helping us to create safer and fairer alternatives for voice assistants.”

Indeed, blockchain is very much part of Snip’s future. As Hindi told TechCrunch in May, the company will release a new set of consumer devices independent of its enterprise business. The idea is to create a consumer business that will prompt further enterprise development. At the same time, they will issue a cryptographic token via an ICO to incentivize developers to improve the Snips platform, as an alternative to using data from consumers. The theory goes that this will put it at odds with the approach used by Google and Amazon, who are constantly criticised for invading our private lives merely to improve their platforms.

As a result Hindi believes that as voice-controlled devices become an increasingly common sight in public spaces, there could be a significant shift in public opinion about how their privacy is being protected.

In an interview conducted last month with TechCrunch, Hindi told me the company’s plans for its new consumer product are well advanced, and will be designed from the beginning to be improved over time using a combination of decentralized machine learning and cryptography.

By using blockchain technology to share data, they will be able to train the network “without ever anybody sending unencrypted data anywhere,” he told me.

And ‘training the network” is where it gets interesting. By issuing a cryptographic token for developers to use, Hindi says they will incentivize devs to work on their platform and process data in a decentralized fashion. They are starting from a good place. He claims they already have 14,000 developers on the platform who will be further incentivized by a token economy.

“Otherwise people have no incentive to process that data in a decentralized fashion, right?” he says.

“We got into blockchain because we’re trying to find a way to get people to participate in decentralized machine learning. We’ve been wanting to get into consumer [devices] for a couple of years but didn’t really figure out the end goal because we had always had this missing element which was: how do you keep making it better over time.”

“This is the main argument for Google and Amazon to pretend that you need to send your data to them, to make the service better. If we can fix this [by using blockchain] then we can offer a real alternative to Alexa that guarantees Privacy by Design,” he says.

“We now have over 14000 developers building for us and that’s really completely organic growth, zero marketing, purely word of mouth, which is really nice because it shows that there’s a very big demand for decentralized voice assistance, effectively.”

It could be a high-risk strategy. Launching a voice-controlled device is one thing. Layering it with applications produced by developed supposedly incentivized by tokens, especially when crypto prices have crashed, is quite another.

It does definitely feel like a moonshot idea, however, and we’ll really only know if Snips can live up to such lofty ideals after the launch.

Ostrichpillow Hood, the latest product from Studio Banana, is no joke

I’m not going to lie, when Studio Banana released the original Ostrichpillow back in 2012, despite breaking all Kickstarter records at the time, I thought the whole thing might be an elaborate joke. Or, worse still, since the sleep-at-your-desk styled product had found popularity amongst people who worked at startups, Silicon Valley was now parodying itself.

Except that “transformative” design company Studio Banana is based in Europe, with offices based in London, Lake Geneva and Madrid. And 500,000 sales and five products later, the joke is arguably on its critics. As I’m fond of telling founders who ask for validation, ultimately it is the market that decides.

Enter the latest Ostrichpillow creation: the aptly named Ostrichpillow Hood. Aptly named because, well, it’s a hood. However, unlike the previous products in the range, which were designed to facilitate sleep in non-traditional places, the Ostrichpillow Hood, we’re told, is to be used in “everyday waking life”.

Specifically, by reducing the ability to see activity in the edges of your field of vision, it is intended to help you focus on the task at hand and/or reduce overstimulation, such as the kind induced by open plan co-working spaces.

The Ostrichpillow Original

“The product we’re launching now is the sixth of the different products that have emerged in the Ostrichpillow family and they’re catering to different needs,” Ali Ganjavian, co-founder of Studio Banana, tells me in a video call yesterday. “Ostrichpillow was really about complete isolation and it was really a statement product… So different products have different use-cases and different functions, and also different social acceptances”.

I suggest that the Ostrichpillow Hood may turn out to be broadly socially acceptable, not least for anyone already familiar with the original Ostrichpillow, but also because asking work colleagues to respect the need to focus is a lot different to asking them to ignore your need to take a nap at your desk. Ganjavian doesn’t degree, even though there is no doubt the two products share the same design heritage.

“A lot of the stuff we are thinking about now is about the state of mind,” he says, noting that throughout the working day we are bombarded with stimuli and information, from messaging apps, emails, social media, meetings and even something as innocuous as having to say hello to work mates. “[The Ostrichpillow Hood] is really about sheltering. It is not only a physical movement, there is psychology in the way it shelters you… it’s about shifting your mood”.

Next Ganjavian demonstrates the three positions the Ostrichpillow Hood is designed to be worn.

The ‘Hood’ position is for when you need to concentrate on something in public, for example when commuting or in an open plan office or coffee shop. Like wearing a pair of visually loud headphones, it also has the added effect of signalling to colleagues that you’d rather not be disturbed or are “wired in“.

The ‘Eclipse’ position, where the hood can be turned around to cover your face completely, is for when you need to truly switch off from your surroundings, such as to deeply think, take a short break or meditate. “If I’ve got my headphones on in that posture then what it allows me to do is to totally isolate myself in the same way I would with an Ostrichpillow but in a much more acceptable way,” says Ganjavian.

Finally, the ‘Hoop’ position, with the hood worn down around your neck, is designed to feel warm and cozy and turns the Ostrichpillow Hood into attire more akin to a fashion accessory.

Adds the Studio Banana co-founder: “What I find really exciting about this moment is that I currently work in between three different geographies, there is so much going on, and how do we create a tool or object that makes me feel good, helps me perform better, and helps me become more efficient, and also feeds that overall well-being that I’m looking for in my workplace. At the same time, I can just walk out into the street with it on and just go home and feel good about it”.

Ultimate.ai nabs $1.3M for a customer service AI focused on non-English markets

For customer service, Ultimate.ai‘s thesis is it’s not humans or AI but humans and AI. The Helsinki- and Berlin-based startup has built an AI-powered suggestion engine that, once trained on clients’ data-sets, is able to provide real-time help to (human) staff dealing with customer queries via chat, email and social channels. So the AI layer is intended to make the humans behind the screens smarter and faster at responding to customer needs — as well as freeing them up from handling basic queries to focus on more complex issues.

AI-fuelled chatbots have fast become a very crowded market, with hundreds of so called ‘conversational AI’ startups all vying to serve the customer service cause.

Ultimate.ai stands out by merit of having focused on non-English language markets, says co-founder and CEO Reetu Kainulainen. This is a consequence of the business being founded in Finland, whose language belongs to a cluster of Eastern and Northern Eurasian languages that are plenty removed from English in sound and grammatical character.

“[We] started with one of the toughest languages in the world,” he tells TechCrunch. “With no available NLP [natural language processing] able to tackle Finnish, we had to build everything in house. To solve the problem, we leveraged state-of-the-art deep neural network technologies.

“Today, our proprietary deep learning algorithms enable us to learn the structure of any language by training on our clients’ customer service data. Core within this is our use of transfer learning, which we use to transfer knowledge between languages and customers, to provide a high-accuracy NLU engine. We grow more accurate the more clients we have and the more agents use our platform.”

Ultimate.ai was founded in November 2016 and launched its first product in summer 2017. It now has more than 25 enterprise clients, including the likes of Zalando, Telia and Finnair. It also touts partnerships with tech giants including SAP, Microsoft, Salesforce and Genesys — integrating with their Contact Center solutions.

“We partner with these players both technically (on client deployments) and commercially (via co-selling). We also list our solution on their Marketplaces,” he notes.

Up to taking in its first seed round now it had raised an angel round of €230k in March 2017, as well as relying on revenue generated by the product as soon as it launched.

The $1.3M seed round is co-led by Holtzbrinck Ventures and Maki.vc.

Kainulainen says one of the “key strengths” of Ultimate.ai’s approach to AI for text-based customer service touch-points is rapid set-up when it comes to ingesting a client’s historical customer logs to train the suggestion system.

“Our proprietary clustering algorithms automatically cluster our customer’s historical data (chat, email, knowledge base) to train our neural network. We can go from millions of lines of unstructured data into a trained deep neural network within a day,” he says.

“Alongside this, our state-of-the-art transfer learning algorithms can seed the AI with very limited data — we have deployed Contact Center automation for enterprise clients with as little as 500 lines of historical conversation.”

Ultimate.ai’s proprietary NLP achieves “state-of-the-art accuracy at 98.6%”, he claims.

It can also make use of what he dubs “semi-supervised learning” to further boost accuracy over time as agents use the tool.

“Finally, we leverage transfer learning to apply a single algorithmic model across all clients, scaling our learnings from client-to-client and constantly improving our solution,” he adds.

On the competitive front, it’s going up against the likes of IBM’s Watson AI. However Kainulainen argues that IBM’s manual tools — which he argues “require large onboarding projects and are limited in languages with no self-learning capabilities” — make that sort of manual approach to chatbot building “unsustainable in the long-term”.

He also contends that many rivals are saddled with “lengthy set-up and heavy maintenance requirements” which makes them “extortionately expensive”.

A closer competitor (in terms of approach) which he namechecks is TC Disrupt battlefield alum Digital Genius. But again they’ve got English language origins — so he flags that as a differentiating factor vs the proprietary NLP at the core of Ultimate.ai’s product (which he claims can handle any language).

“It is very difficult to scale out of English to other languages,” he argues. “It also uneconomical to rebuild your architecture to serve multi-language scenarios. Out of necessity, we have been language-agnostic since day one.”

“Our technology and team is tailored to the customer service problem; generic conversational AI tools cannot compete,” he adds. “Within this, we are a full package for enterprises. We provide a complete AI platform, from automation to augmentation, as well as omnichannel capabilities across Chat, Email and Social. Languages are also a key technical strength, enabling our clients to serve their customers wherever they may be.”

The multi-language architecture is not the only claimed differentiator, either.

Kainulainen points to the team’s mission as another key factor on that front, saying: “We want to transform how people work in customer service. It’s not about building a simple FAQ bot, it’s about deeply understanding how the division and the people work and building tools to empower them. For us, it’s not Superagent vs. Botman, it’s Superagent + Botman.”

So it’s not trying to suggest that AI should replace your entire customers service team but rather enhance your in house humans.

Asked what the AI can’t do well, he says this boils down to interactions that are transactional vs relational — with the former category meshing well with automation, but the latter (aka interactions that require emotional engagement and/or complex thought) definitely not something to attempt to automate away.

“Transactional cases are mechanical and AI is good at mechanical. The customer knows what they want (a specific query or action) and so can frame their request clearly. It’s a simple, in-and-out case. Full automation can be powerful here,” he says. “Relational cases are more frequent, more human and more complex. They can require empathy, persuasion and complex thought. Sometimes a customer doesn’t know what the problem is — “it’s just not working”.

“Other times are sales opportunities, which businesses definitely don’t want to automate away (AI isn’t great at persuasion). And some specific industries, e.g. emergency services, see the human response as so vital that they refuse automation entirely. In all of these situations, AI which augments people, rather than replaces, is most effective.

“We see work in customer service being transformed over the next decade. As automation of simple requests becomes the status-quo, businesses will increasingly differentiate through the quality of their human-touch. Customer service will become less labour intensive, higher skilled work. We try and imagine what tools will power this workforce of tomorrow and build them, today.”

On the ethics front, he says customers are always told when they are transferred to a human agent — though that agent will still be receiving AI support (i.e. in the form of suggested replies to help “bolster their speed and quality”) behind the scenes.

Ultimate.ai’s customers define cases they’d prefer an agent to handle — for instance where there may be a sales opportunity.

“In these cases, the AI may gather some pre-qualifying customer information to speed up the agent handle time. Human agents are also brought in for complex cases where the AI has had difficulty understanding the customer query, based on a set confidence threshold,” he adds.

Kainulainen says the seed funding will be used to enhance the scalability of the product, with investments going into its AI clustering system.

The team will also be targeting underserved language markets to chase scale — “focusing heavily on the Nordics and DACH [Germany, Austria, Switzerland]”.

“We are building out our teams across Berlin and Helsinki. We will be working closely with our partners – SAP, Microsoft, Salesforce and Genesys — to further this vision,” he adds. 

Commenting on the funding in a statement, Jasper Masemann, investment manager at Holtzbrinck Ventures, added: “The customer service industry is a huge market and one of the world’s largest employers. Ultimate.ai addresses the main industry challenges of inefficiency, quality control and high people turnover with latest advancements in deep learning and human machine hybrid models. The results and customer feedback are the best I have seen, which makes me very confident the team can become a forerunner in this space.”

Ola raises $50M at a $4.3B valuation from two Chinese funds

Ola, the arch-rival of Uber in India, has raised $50 million at a valuation of about $4.3 billion from Sailing Capital, a Hong Kong-based private equity firm, and the China-Eurasian Economic Cooperation Fund (CEECF), a state-backed Chinese fund. The funding was disclosed in regulatory documents sourced by Paper.vc and reviewed by Indian financial publication Mint.

According to Mint, Sailing Capital and CEECF will hold a combined stake of more than 1% in Ola . An Ola spokesperson said the company has no comment.

Ola’s last funding announcement was in October, when it raised $1.1 billion (its largest funding round to date) from Tencent and returning investor SoftBank Group. Ola also said it planned to raise an additional $1 billion from other investors that would take the round’s final amount to about $2.1 billion.

At the time, a source with knowledge of the deal told TechCrunch that Ola was headed toward a post-money valuation of $7 billion once the $2.1 bllion raise was finalized. So while the funding from Sailing Capital and CEECF brings it closer to its funding goal, the latest valuation of $4.3 billion is still lower than the projected amount.

Ola needs plenty of cash to fuel its ambitious expansion both within and outside of India. In addition to ride hailing, Ola got back into the food delivery game at the end of last year by acquiring Foodpanda’s Indian operations to compete with UberEats, Swiggy, Zomato and Google’s Areo. It was a bold move to make as India’s food delivery industry consolidated, especially since Ola had previously launched a food delivery service that shut down after less than one year. To ensure the survival of Foodpanda, Ola poured $200 million into its new acquisition.

A few months later after buying Foodpanda, Ola announced the acquisition of public transportation ticketing startup Ridlr in an all-stock deal. Outside of India, Ola has been focused on a series of international launches. It announced today that it will begin operating in New Zealand, fast on the heels of launches in the United Kingdom and Australia (its first country outside of India) this year.

Congressional bill would improve startup valuations

Late last week, Congress moved one step closer to passing the American Innovation Act of 2018, a bill that would make accounting and tax changes that would likely increase the valuation of startups in an acquisition.

The House Ways and Means committee approved a bill containing text that would improve the treatment of Net Operating Losses (NOLs) for startups. While many startup founders would probably rather watch paint dry (or build their companies) than dive into complex tax code changes, the provisions in the bill could greatly improve the ability of startups to invest in growth activity, and could drive meaningfully positive impacts to valuations, acquisition prices, capital markets participation and venture returns.

First, though, what are NOLs? Each year, if a company loses money, it can claim the losses as a deduction off of its future taxes. Traditionally, the U.S. tax code has allowed companies to cumulatively track and carry forward NOLs to offset taxable income in future years, reducing the amount of cash required to pay taxes. These NOLs are essentially a cash-like asset, and they can be exchanged in the event that a company is acquired.

However, a long-standing IRS provision, Section 382, which was originally implemented to prevent companies with large tax appetites from acquiring those with large operating losses exclusively to reduce taxes, limits the use of NOL carry-forwards in instances of ownership change. 

Currently, in cases of an ownership change, specified as a more than 50 percent change in the ownership of shareholders who own at least 5 percent of a company’s stock, the amount of taxable income for the “post-change” company that can be offset by existing NOLs cannot exceed the value of the “pre-change” company, multiplied by the long-term tax exempt rate set by the IRS.

(Yes, this is why you hire a tax attorney.)

The net-net is that this provision has been particularly challenging for startups, which often trigger this limiting condition, given they frequently operate in the red through growth stages and often see frequent, sizable changes in their ownership structure due to fundraising, public offerings and acquisitions.

The House bill would alleviate this complication by protecting these tax offsets and creating an exception to the section 382 provision for startups, allowing the application of NOLs and R&D tax credits realized in the first three years of operations regardless of ownership change limitations.

These changes have a number of benefits for startups. It would provide increased flexibility around early-stage financing activities and remove potential issues that could arise with capital markets activity. Additionally, with startups more easily maintaining tax offsets to reduce their cash taxes, startups would have larger cash balances to invest in growth efforts.

The protection of the NOL from ownership change limitations could also have serious impacts to company valuations and the attractiveness of startups as acquisition candidates. With acquirers better able to utilize existing tax offsets, startups should benefit from higher purchase prices from the inclusion of NOL balances in valuations, helping founder and VC returns.

The bill passed through committee through a voice vote with no objections and is now expected to be voted on by the rest of the House later this month before advancing to the Senate. The bill has 23 co-sponsors, all Republican.

Mumford & Sons beware! An AI can now write indie music

A fascinating project called Amadeus Code promises to out-Tay-Tay Tay Tay and out-Bon Bon Iver. The AI-based system uses data from previous musical hits to create entirely new compositions on the fly and darn if these crazy robots-songs aren’t pretty good.

The app, which is available from the iTunes Store but doesn’t seem to be working properly, creates song sketches in minutes, freeing you up to create beautiful lyrics and a bit of accordion accompaniment.

The video above is a MIDI version of an AI-produced song and the video below shows the song full produced using non-AI human musicians. The results, while a little odd, are very impressive.

Jun Inoue, Gyo Kitagawa, and Taishi Fukuyama created Amadeus Code and all have experience in music and music production. Inoue is a renowned Japanese music producer and he has sold 10 million singles. Fukuyama worked at Echo Next and launched the first Music Hack Day in Tokyo. Fukujama is the director of the Hit Song Research Lab and went to Berklee College of Music.

“We have analyzed decades of contemporary songs and classical music, songs of economic and/or social impact, and have created a proprietary songwriting technology that is specialized to create top line melodies of songs. We have recently released Harmony Library, which gives users direct access to the songs that power the songwriting AI for Amadeus Code,” said Inoue. “We uniquely specialize in creating top line melodies for songs that can be a source of high quality inspiration for music professionals. We also do have plans that may overlap with other music AI companies in the market today in terms of offering hobbyists a service to quickly create completed audio tracks.”

When asked if AI will ever replace his favorite musicians, folks like Michael & Janet Jackson or George Gershwin, Inoue laughed.

“Absolutely not. This AI will not tell you about its struggles and illuminate your inner wolds through real human storytelling, which is ultimately what makes music so intimate and compelling. Similarly to how the sampler, drum machine, multitrack recorder and many other creative technologies have done in the past, we see AI to be a creative tool for artists to push the boundaries of popular music. When these AI tools eventually find their place in the right creative hands, it will have the potential to create a new entire economy of opportunities,” he said.

Boom’s chief test pilot on the thrill and challenge of going supersonic (again)

“There’s nothing like it out there,” says Retired Commander Bill “Doc” Shoemaker, chief test pilot for Boom Supersonic, the startup aiming to make a passenger airliner for transoceanic flights at speeds (as you might guess from the name) faster than sound. Shoemaker, a former Navy aviator, fighter pilot, and aeronautics engineer, will have the daunting privilege of being the first to fly the company’s proof of concept single-seater during tests next year.

That there’s nothing like Boom is not exactly a controversial opinion — there aren’t a lot of companies out there trying to resurrect supersonic flight. The Concorde is, after all, so well known a cautionary tale of engineering ambition exceeding the constraints of reality that it verges on hackneyed. But Shoemaker isn’t a silicon valley startup commentator, he’s a test pilot, and his perspective is that of someone who has worked on and flown dozens of aircraft, including supersonic ones, over his decades-long career.

The first question I asked (though not entirely a serious one) when I had a chance to chat with Shoemaker was whether it was a bit premature to have a chief pilot at a company that doesn’t yet have a plane to fly.

“There’s a good reason to have a pilot at this point,” Shoemaker said. As he delicately put it: “Among the team, the pilots are… uniquely committed to the outcome.”

Among other things, test pilots seem to have a knack for understatement. But it’s certainly true.

“You want the operator’s perspective, like how to build the cockpit, how you’ll operate the aircraft. The designer will come to me for that perspective — he’ll say, ‘how can I tweak the design to be more suitable for you?’ You want that cross-industry expertise.”

Boom is making a supersonic airliner, but it’s still mostly a paper plane, if you will. The company’s test craft, the XB-1, however, is being built and should be taking to the air about a year from now. That’s where many of the components, materials, and design choices will be flight-proven. Interestingly, however, actually flying the test craft is a rather analog affair.

“The aircraft is definitely designed around a philosophy, which ‘keep it simple.’ We’re not trying to introduce any more tech than we really need to. The flight controls are not fly-by wire, they’re mechanical,” explained Shoemaker. “It’s going to be an interesting airplane to fly. It goes from 150 knots up to Mach 2.2, and up to 45,000 feet. It’ll be a challenge because of that mechanical stuff, but with what we’re trying to do, keeping it simple makes a lot of sense.”

That’s not to say nothing has changed over the last few decades of aeronautics, a topic in which, if you’ll recall, Shoemaker has a doctorate. Although he said he considers his role as being separate from the flight test engineers who put the craft he’s flown together, he’s still an important part of the team.

He suggested a few areas where he’s seen or expects improvements to the aircraft creation and testing process.

“One is composite materials. That’s huge,” he said, referring to things like carbon fiber and more exotic weaves and alloys that combine a number of desirable characteristics. “The strength and weight improvements offer new opportunities. You know, the Concorde would contract like a foot during flight temperatures, then expand again. Composites don’t do that. All these things make the aircraft lighter, faster, and stronger.”

Second, he briefly noted, engine technology these days is “brisk,” especially combined with the materials advances.

“Last,” he said, “the Concorde design was wind tunnel based, but a lot of the work we do is computation. We can do all the testing they did for the Concorde in a couple days.”

Wind tunnels are still involved, of course, but the models are so good that it’s more for verification than testing. But it also lets designers speed through ideas, evaluating but skipping wild ones without wasting time: “You can look at all these weird corner cases, and explore those very quickly.”

Basic advances in tech mean the team can avoid quirks like the Concorde’s drooping nose, which was there so that pilots could see the runway. “You can all the mechanical complexity that comes with that,” said Shoemaker. “For us we’ll be going with a direct camera or some kind of vision system that’s integrated with all the systems.”

“The airliner itself,” he said, “will be highly augmented [compared to the test jet]. It’ll be fly by wire. Its handling qualities are really quite benign across the envelope. It’s surprising but the way the aircraft handles on one side of the speed of sound isn’t so different from how it handles on the other side.”

Ultimately Shoemaker was optimistic about the whole enterprise, both the company and the prospect of supersonic passenger flight.

“As far as an ambitious project with an ambitious goal, there’s nothing like it out there,” he said. “That’s the value and reward of working with a team this size, a team that really believes they can reinvent and do it better. And it’s well within what we can do with technology — we can do it better than Concorde did, possibly by orders of magnitude.”

As for his part, the test flights set to take place next year, he’s more than a little excited.

“It’ll be a challenge to fly for sure — but it’ll be nice to go that fast again.”

Blippar picks up $37 million hoping to become profitable in the next year

Blippar, the AR startup that launched in 2011, has today announced the close of a $37 million financing led by Candy Ventures and Qualcomm Ventures.

The company started out by offering AR experiences for brand marketers through publishers and other real-world products, letting users unlock AR content by scanning a tag called a “Blipp”.

Blippar then transitioned to a number of different AR products, but took a particular focus on computer vision, launching a consumer-facing visual search engine that would let users identify cars, plants, and other real-world objects.

Most recently, Blippar has introduced an indoor positioning system that lets commercial real estate owners implement AR mapping and other content from within their buildings.

The AR industry has been in a state of evolution for the past few years, and Blippar has constantly reshifted and re-positioned to try and take advantage of the blossoming market. Unfortunately, several pivots have put the company in a tough spot financially.

BI reports that Blippar posted revenue of £8.5 million ($11.2 million) in the 16-month period up to March 31 2016, with losses of £24 million ($31.5 million). These latest rounds have essentially let Blippar keep the lights on while trying to pick up the pace on revenues.

The company says that this latest round is meant to fuel the company’s race to reach profitability in the next 12 months. Blippar has raised more than $137 million to date.