Shapearrow_downarrowleft_whitearrowright_whitesubmit-iconShapeShapeShapeicon_facebookicon_facebookShapeGroup 2success-iconCombined ShapeShapeShapeglyph-logo_May2016glyph-logo_May2016interneticon_linkedinicon_linkedinGroup 10Group 2ShapeShapeShapeShapeTriangle 1facebooklinkedintwitterxingShapesnap-ghostsnap-ghostShapetwitterShapeShapeShapeShapeShapeShapeGroup 5Group-2Xing

Artificial Intelligence (AI) is no longer Science Fiction, no longer only a vision or a potential future threat. AI has become applicable, even a buzzword these days and is still one of the most promising areas, spinning around in the startup ecosystems of the world.

It seems we are flooded by news about AI. Every day there are new articles about breakthroughs by young, well-funded startups or new funding records, led by tech giants as Google or Apple. Not long ago, AI was only a concept, a dream of scientists. Today, we permanently carry devices with us that have more computer power and are more intelligent that the computers that were used for the first moon landing. Consciously or not, AI takes already place in our daily life.

In this paper, we will explore the current wave of AI businesses and the technologies behind it. We will give a profound overview of the current market activities related to Artificial Intelligence (AI), and discuss the implications for the financial- and health sector as well as the development of smart home applications.

Global Key Figures

€ 2.173 Mio
Disclosed investments into AI in 2015
397
AI equity deals in 2015
€ 4,6 BN
Market for Artificial Intelligence in 2020
53,65%
AI market growth till 2020 (CAGR)
Test it

Chapter

The new wave of Artificial Intelligence

Big Data, Algorithms and Computing Power

Abstract

One of the great, current meta-themes is Artificial Intelligence. But what do we mean by this? What is the state of development? And how can it be applied?

1 Enabling the potential of AI

Besides the new wave of Artificial Intelligence that came up in the last years, AI has been there for decades. People always aimed to develop machines that take over human tasks. Already in 1950 Turin published a paper discussing the dissemination of AI, which only existed as a concept, not yet as a real application. This changed in the 1980´s where Artificial Intelligence developed drastically. IBM`S Deep Blue won chess again world champion Gary Kasparov, other functionalities were tested and proved by the military, as the Predator unmanned aerial vehicle during the Balkan War. From 2005 onwards, large tech companies such as Google, IBM or Facebook have invested heavily into AI and have paved the way for more commercial application. Nowadays AI solutions are able to keep up with humans in specific areas. DeepMind succeeded in March to beat one of the world’s top masters in the Asian board game Go, which was previously considered too complex for computers. The machine surprised at one point even experts with a creative train, to date no one had ever played. Why did the new wave of AI arise? What changed within the last 20-30 years, that AI has become that present?

The rise of AI was enabled by the availability of vast amounts of data, enabling computers to improve themselves by finding patterns and actively learn, the explosion of computing power and improved algorithms which facilitated the existence of intelligent systems.

With the introduction of Apple´s Siri in 2011, it was easy to see how little intelligent a system without adequate training by large amounts of data can be. In Amazon’s launch of Echo in 2015, it was clearly recognizable which leaps were made in this area in the last four years. Even though there are still steps to take, in order to create a truly intelligent system, personal assistants like Siri, Cortana, Echo, ViV or the cognitive system IBM Watson are first steps to make Artificial Intelligence more visible, approved and accepted by consumers.

The coexistence of Big Data and Artificial Intelligence, which adds and intelligent layer on the generated data, enables us to analyze and understand complex data and interrelations much faster and efficient than humans could ever do. And the initial position regarding the availability of data is quite promising!  90% of the world´s data has been generated in the last two years. Annual global IP traffic will surpass the zettabyte threshold in 2016. By the end of 2020, there will be more than 212 billion connected devices around the world. Global Internet traffic in 2020 will at least be equivalent to 95 times the entire internet volume of 2015.

However, not least, improved algorithms and deep learning as well as the availability of affordable parallel processing accelerate the expansion of AI. For instance, clusters of modern graphic processors can compute immediately, what traditional processors needed days for. Prisma, an AI app that can repaint photos to look like it was composed by a famous artist, recently showed how advanced the state of technology is.

Consequently, it is not surprising that the market for Artificial Intelligence solutions is predicted to explode from initial €377,7 million in 2015 to €4,6 billion in 2020. This equals a CAGR of 53,65% between 2015 and 2020. The market for enterprise AI systems will grow from €182,3 million in 2015 to €9,9 billion by 2024. That’s even a 56% compounded annual growth. Worth to notice, 72% of all AI-Startups focus on a B2B business model.

This however, can be attributed to the fact that, while the amount of data increases exponentially, most of it is collected by major corporations and other institutions and not publicly available. Thus, startups need to work together with these larger companies to access these datasets or create ways that motivate individuals to generate the data that is required for the AI to learn, as Hendrik Makait, Data Scientist at Project A Ventures, says.

2 What is AI?

There is still no proper consensus about the definition of Artificial Intelligence, as there is barely a common standard definition for intelligence. Following Google DeepMind, relying on the definition of the Encyclopedia Britannica, Artificial Intelligence, „ is the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. The term is frequently applied to the project of developing systems endowed with the intellectual processes characteristics of humans, such as the ability to reason, discover meaning, generalize or learn from the past experience,”[1] whereat Artificial Intelligence is generally clustered into two approaches, strong and weak AI.

Strong AI, creates intelligence that thinks like humans and can solve problems, including awareness, self-confidence and emotions, e.g. DeepMind. Weak AI focuses on applications of interest to the solutions to which generally a form of “intelligence” is required, including the simulation of intelligent behavior by means of mathematics and computer science, e.g. most chatbots or Apple´s Siri.

There is another distinction to be made here, when talking about Startups, working with AI. One has to differentiate between AI systems designed for specific tasks known as narrow AI, and those systems that are designed for the ability to reason in general, general AI.

[1] Microsoft’s Cortana and Apple´s Siri are therefore not truly intelligent, because they can only respond to pre-programmed situations.

3 Applications of AI

Due to the fact, that there is such a debate about the definition, and the different approaches, it is of avail, to cluster different field of application when talking about Artificial Intelligence, as it appears in many different forms on various applicable verticals. We looked at more than thousand startups an built the following cluster:

  1. Machine Learning/Deep Learning, which represents 34% of all AI companies. Startups in this field focus on the utilization of computer algorithms that work in vertically specific use cases, depending on the generated data, e.g. TwentyBN.
  2. Natural Language Processing, which represents 21% of all AI companies. Startups in this field process sound clips of human speech as wel as texts and derive meaning from them, e.g. Verbio or VoiceBase.
  3. Computer Vision & Image Recognition, which represents 17% of all AI companies. Startups in this field focus on the utilization of technology that processes and analyses images in specific use case in order to recognize objects, e.g. Flyby, Cortica or Leverton.
  4. Virtual Personal Assistants, which represents 8,2% of all AI companies. Startups in this field work on intelligent personal assistants that perform everyday tasks and services, e.g. Siri or Vlingo.
  5. Smart Robots, which represents 5,8% of all AI companies. Startups in this field work on smart robots that can act and react autonomously, e.g. Jibo or Anki.
  6. Recommendation Engines and Collaborative Filtering which represents 5,4% of all AI companies. Startups in this field work on software, that predicts the preferences and interest of users, e.g. Fligoo or Snapsort.
  7. Gesture Control, which represents 3% of all AI companies. Startups in this field work on the uninterrupted communication with computers through gesture, e.g. Zenglove.
  8. Content Aware Computing, which represents 2,5% of all AI companies. Startups in this field work on software that is aware of its environment and its context of use, e.g. Clever Sense.
  9. Video Automatic Content Recognition, which represents 1,3% of all AI companies. Startups in this field work on software sampling of video content with a source, e.g. Cognitive Networks.
  10. Speech to Speech Translation, which represents 1,26% of all AI companies. Startups in this field work on software that, automatically and instantly, recognizes and translates human speech, e.g. Apptek.

 

Insert your data to continue reading
Subscribe

Chapter

Investors rush into AI

Characteristics and Investments into AI

Abstract

A new record in AI financing can be recorded for 2015, and the trend towards an upshift is even more promising in 2016. What does an AI Startup look like, and how about the funding and Investor´s landscape? The following will bring some valuable insights.

1 Demographics

Let’s have a look at the demographics of AI startups worldwide. For how long are they revolutionizing the world, and where do they actually work on it?

Age

Startups working in the field of Artificial Intelligence exist on average for 7,25 years. It is obvious that many of them just got founded around the creation of new research successes and technological advancements. However, the average age of AI startups varies strongly across the different categories. While some areas, especially those related to recognition and translation are already more than eight years old on average, quite young startups mainly focused on Machine Learning, AI based Recommendations Engines or Context Aware computing, only exist for a few years on average.

Location

A view on the location, reveals a high unbalance, in terms of startup density, between the USA and Europe. Approximately 64% of all AI startups are based in the USA. The country with the most startups in Europe, the United Kingdom, contributes only 9% to the worldwide appearance – , Germany 4% and France roughly 3%. A narrower view at the location shows, that remarkable 4,7% percent of global AI Startups have their office in the Silicon Valley, 1,8% in New York, only 0,8% in London and only 0,6% in Berlin. While there is a strong cluster in the SF Bay Area, with 8,1% of all identified AI Startups, In particular Germany leaves space for improvements in terms of startup density. Northern America clearly dominates the global AI landscape, not only due to a higher density of companies in the US but also regarding research. Canada for example, is one of the global hotspots for AI, as Dr. Christian Thurau, CTO of Twenty BN, a German AI startup with a research site in Toronto and a strong network into the AI scene, says. The good position, especially in the US, also stems from the well-developed investment and research environment, which is traditionally more active here.

 

2 Investment Landscape

Investments

Not only Twenty BN, who besides their research site in Canada, recently received their seed funding from a business angel, and AI expert in New York City, also other German startups such as the Munich-based, smart sensor developer Konux state, that the conditions in Germany are not stunning. Konux has made a conscious decision to search for potential investors in the US, where more Venture Capital is dedicated to AI.

Accel, the early- and growth stage investor, with its headquarter based in Palo Alto, is among the most active VC investors into AI, with 23 investments, followed by NEA, as well based in Palo Alto, with 18 investments. In Germany, there are only a few funds, actively investing into AI, as Fabian Westerheide, Managing Director of Asgard, a German VC specialized on AI, says.

Besides many other VCs, focused or generic, it is noteworthy, that more and more CVCs are among the most active investors into AI Startups. Ranked first worldwide is Intel Capital, who invested in more than a dozen AI focused Startups. A trend that can be observed, not only since Mark Zuckerberg announced the development of his own intelligent personal assistant, is that many companies working on Artificial Intelligence have been acquired by big tech companies such as IBM, Facebook, Google or Apple in the last years.

There is an intense competition between the tech giants to acquire the most promising startups with the most advanced technologies. In 2016, there have already been six major acquisitions. Google has emerged to become the most active tech-acquirer in the last five years with more than 10 acquisitions. Most recently, the internet giant announced that it bought the French Startup Mood Stock, whose  technology can recognize objects with a smartphone. Google also invested into the German Research Center for Artificial Intelligence (DFKI) in Saarbrücken, striving to ensure access to new technologies with great commercial potential. The DFKI is closely cooperating with German startups, such as Leverton, a service provider specialized in intelligent information extraction from contracts. Intel just bought the two year old, deep learning startup Nervana Systems for $350.

Overall, there have already been more than 20 acquisition by big tech companies this year. Behind Google, Twitter ranks second with four acquisitions, including the acquisitions of TellApart and Whetlab. Apple, Salesforce, AOL, IBM and Yahoo rank third, whereas Apple and Salesforce just joined the race recently. However, at least 60% of the 30 companies that were acquired by the big tech corporates in the last five years had Venture Capital backing. Most AI Startups were acquired already four years after their first funding.[2] M&A activities increased  seven times  from only five transactions, five years ago to 35 in 2015.

There is not only competition and traction on the investment side. The large IT and Internet companies have systematically managed to absorb the super talents of science and research. Featuring lush research budgets and highly motivated teams of experts, Google, Facebook or Apple are working feverishly to exploit the potentials of Artificial Intelligence for themselves increasingly shifting research from universities towards corporates as Christian Thurau experienced. This might consequently lead to a bait ofGerman experts.

Investment Volumes

Investment volumes are rapidly increasing. Regarding current data from CB insights, Q´1 16 has marked a new long term high in global AI equity deals, with 134 deals including investments in companies like Trifacta, Indigo or Pathway Genoics. Though the amount of deals fell to 121 in Q´2 16 funding volumes reached a new long-term high with € 954.56 Mio. The number of deals in Artificial Intelligence have increased by approximately six times from 2011 to 2015. Funding Volumes have increased by more than eight times, from € 256.6Mio in 2011 to € 2.173 Mio in 2015. This equals a 53% compounded annual growth in funding volumes.[3]

Funding by Category

Having a look at the funding landscape per AI category, one can see that on average, Machine Learning gets the highest fundings, followed by Computer Vision. Natural Language Processing takes in the second place regarding the total funding volume. Gesture Control and Smart Robots receive lower total funding.

The few startups working in this field receive on Average nearly 50% of what machine learning companies raise.

Only 20 AI focused companies raised Series D rounds or beyond between 2011 and 2015.

[2] E.g. Emu, acquired by Google after only one year after Seed funding, Tempo AI which was acquired by Salesforce Series A.

[3] The huge number of disclosed investments in Q´4 14 is due to the closing of four mega rounds.

Chapter

Use Cases

Financial Services, Digital Health, Smart Home

Abstract

Businesses begin to use the potentials of Artificial Intelligence. There are some outstanding applications across varies industries.

1 Financial Services

Technology always had a big share in modern financial services. Fraud detection or stock trading are based on IT . Most banks are already software companies – AI will improve efficiency even further.

Goldman Sachs, Morgan Stanley, or UBS, all oft the big financial institutes are exploring the use and potential of Artificial Intelligence software. Goldman Sachs let a $15 Mio investment into Kensho, which combines natural language search queries, graphical user interfaces, and secure cloud computing to create analytics tools for investment professionals. Charles Schwab, the American brokerage and banking company, has recently established its own fully automated, AI based investment advisory service. Bridgewater, the world largest hedge fund established its own AI team in 2015, with about 6 experts creating trading algorithms for market predictions.

Even German institutes start to recognize the power of AI. This year, Deutsche Bank announced its crowdstorming AI concept, inviting people to submit their concepts on AI. The bank hopes to be at the forefront of AI application in financial services. In addition, the Main Incubator, part of the Commerzbank Group made a strategic investment into GINI, a B2B AI startup which offers semantic document analysis.

However, there are various applications of Artificial Intelligence in financial services, including for instance, the so called Reg-Tech. Regulations proliferated after the financial crisis in 2007. Finance companies increasingly have to stick to guidelines such as the European Market Infrastructure Regulation, in Europe or the Dodd-Frank Act in the US.[4] Many experts had to be hired, pushing up the cost position of most corporations. Artificial Intelligence software nowadays, is steadily supporting those employees, as modern algorithms are capable of taking care of anti-money laundering programs, “know your customer checks” or billing fraud oversights. Artificial Intelligence makes those processes more efficient, reduces costs and can improve overall quality – AI enables those institutes to use its recourses more efficient

Another AI application is Personal Financial Management, which due to an increased degree of automation, can be more customized real-time services at low costs. Smart Wallets, for instance, are capable of analyzing and optimizing the user´s consumer habits in real time. One company enabling this optimization is the US based Startup Wallet.AI. which according to founder and CEO, Omar Green, “builds machines to help consumers make smarter decisions about their money, especially when they’re out spending it”.

Other AI driven approaches include the optimization of lending decisions for creditors, wealth management advisory or other automated financial advisory services such as the German startup Scalable Capital, a digital investment service or Vaamo who recently announced their cooperation for investment services with N26.

There are even promising fields of application within the insurance industry – Automating the underwriting process, facilitating a better decision-making process or automating claim management. In times, where huge amounts of data are generated through wearables and other devices, Artificial Intelligence will be the smart layer on this data – making Big Data smart is the strategic mean for insurance companies.

AI already has its place in financial services. Most stock exchanges, for example, are already a machine’s world.

[4] Deloitte estimated, that the European insurance industry spent between $5.7 and $6.6 billion in 2012 to comply with new regulations.

It is core and vital to the business. We like to say we’re a technology company that happens to be in finance.

Dustin Lucien, CTO of Betterment

It's lower cost for the investor. As opposed to working with a traditional advisor where you might pay up to 1%, here you get portfolio management at essentially no management fee.

Tobin McDaniel, President of Schwab Wealth Investment Advisory, Inc

We use real-time Artificial Intelligence to rate the likelihood of card misuse, thus reducing the risk of fraud. Artificial intelligence means that the system continually gets smarter based on each successful instance of fraud detection.

Christian Rebernik, CTO of N26

2 Digital Health

The availability of extensive data in the health industry is a remarkable prerequisite for the usage of Artificial Intelligence solutions, which enables the simplification of data collection, evaluation and the deduction of suitable decisions while decreasing the costs by half.

According to consulting firm Frost & Sullivan, AI in health care will face a dramatic market expansion. “By 2025, AI systems could be involved in everything from population health management to digital avatars capable of answering specific patient queries”. Possible fields of application include Predictive Care, Image Interpretation, or Diagnostic Support. Through Artificial Intelligence implementation, early diagnosis can be improved, correlations and influencing factors can be analyzed and preventions and early diagnosis can increase in efficiency.One startup working on this application is the Berlin-based company

One startup working on this application is the Berlin-based company Heurolabs, whose algorithm is able to recognize malignant tumors on images through independent learning, based on a vast amount of data, pictures of tumors. By this technology, whenever diagnoses get complex, the Heurolab algorithm can be used as supporting for any decision the doctor might be uncertain about.

Another good example how Artificial Intelligence will influence health care and predictive care is the startup Moodstock. Among other applications, the software intends to help in the detection of eye diseases, the scans are currently still evaluated by doctors. The British Moorfields Eye Hospital will make more than a million eye scans anonymized available for the software, which is looking for signs of disease.

Additionally, innovative services and projects are increasingly using the cognitive system of IBM Watson. The American company CVS Health cooperates with IBM to offer preventive health services to clients with chronical illnesses. Watson will contribute decisively to draw conclusions from patient’s data about a possible deterioration of health. Furthermore, researchers from the University of Cambridge have developed an application based on Watson, that shows users possible treatments of health problems. Users can send in a description of the symptoms, and the app returns a suggested diagnoses and treatment options.

The huge potential of Artificial Intelligence in healthcare is also reflected in the investment landscape. Deals into AI Companies, have steadily increased from 2011 onwards, reaching a new high Q1´16 with 30 deals, increasing more than 7 times from 2011 to 2015. Healthcare was responsible for 15% of all AI deals in 2015, whereas most health care related AI startups got Early Stage investments (46% Seed/Angel, 23% Series A).

According to Prof. Jürgen Schmidhuber, scientific director of the Swiss AI Lab IDSIA, automation will decrease costs and therefore make medical diagnosis accessible for a wider range of people. In that case, one of the most important assets that medical institutions have nowadays is their data. Especially in Germany investors and startups still have to cope with strict regulations. However, regulation follows innovation.

The world spends over 10% of GDP on healthcare (over 7 trillion USD per year), much of it on medical diagnosis through expensive experts. Partial automation of this could not only save billions of dollars, but also make expert diagnostics accessible to many who currently cannot afford it. In this context, the most valuable asset of hospitals may be their data – that’s why IBM spent a billion on a company that collected such data.

Prof. Jürgen Schmidhuber, Scientific Director of The Swiss AI Lab IDSIA

Digital Health is a very interesting field of application for AI , unfortunately , it is also highly regulated.

Fabian Westerheide, MD of Asgard Capital

3 Smart Home

Welcome to Jarvis! At the beginning of this year, Mark Zuckerberg the Facebook CEO announced, that he wants to build an AI for his home that is like Iron Man´s Jarvis, an intelligent personal assistant (IPA) that can manage his house and can present virtual reality presentations. Unfortunately, this product seems way more advanced than what is in the market right now.

Besides some Virtual Personal Assistants, who are capable of playing music or dimming the light, and some smart thermostats who regulate the room temperature and learn a certain behavior, the current market still is some steps away from a fully intelligent home. Several advancements regarding Virtual Personal Assistants such as the recently announced ViV, or Amazon´s echo are indicators for where the journey goes. Those IPAs will also integrate the numerous components for smart homes that are already available, such as flexible lighting, automated shutters or electronic gadgets such as the tv or the coffee machine. Home Service Robots could be the next breakthrough, serving meals, cleaning the house , taking care of the elderlies or reading out bed-time stories. The social robot Jibo indicates how a robot could be like.

The main driver in Germany will in particular be the issue of energy efficiency and demographic change. A steady increase in energy costs will push the demand for more intelligent energy efficiency management and adjustable home automation. Intelligent technical assistance will increasingly help to satisfy the desire for old-age, independent living in one own home. Especially innovations from B2B applications, where the whole value chain could be digitized, will find their ways into private households.

The German Smart Home Market alone will increase to over €19 billion by 2025. Artificial Intelligence will accelerate this development as it creates more value, comfort and convenience for customers.

"My personal challenge for 2016 is to build a simple AI to run my home and help me with my work. You can think of it kind of like Jarvis in Iron Man."

Mark Zuckerberg, CEO of Facebook

Chapter

What to do with the new Intelligence?

Abstract

Sections

In the near future, the exponential improvement of convolutional neural networks will continue, especially in conjunction with significant computing power and supercomputers. Although there is a long way to go, computer will improve their emotional understanding, leading to a virtually uninterrupted interaction between human and computers, also amplified by better cameras or improved facial and voice recognition.

Through an increasing combination and link of technical devices, and thus more need for AI, the question will arise, how to integrate AI into our daily lives. Especially, when AI will no longer only be an interface between people, partly as a co-worker, ethical questions need to be clarified.

There are significant potentials employing AI technology and anyway no turning back. An increase in efficiency through lower error rates and more precision, reduction of costs, and the reduction of emotional barriers, such as stress are only some mentionable points, accelerating the automation of financial services, the health industry and smart homes. However, the technology also brings various issues. Jobs are predicted to be eliminated, security, business and privacy questions arise.

It will be important, that intelligent programs are developed in a way, that secures the desired benefits, including a detection of harmful influences. Political, social and ethical frameworks need to be discussed, updated and improve, Artificial intelligence comprises both, opportunities and threats.

However, as Sundar Pichai, the CEO of Google, recently said, “ we will move from mobile first to an AI first world”. That’s for sure.

Christian Renner

Managing Director of Kompass Digital

Carl-Luis Rieger

Investment Analyst at Kompass Digital

Theodor Schulte

Business Analyst at Kompass Digital  

About Kompass

Kompass Digital is a venture capital fund located in Berlin. Through early stage investments, we aim to back digital, EU-based startups with the potential to shape industries and expand their reach globally. We consider lead and co-investments in the following areas: Financial Services, Digital Health and Smart Home. We strongly believe that startups should do what they do best - build a company. We are here to support entrepreneurs on that journey.