Neuromation Q2 Quarterly Report

Contents:

• Letter from the CEO
• Introduction
• Neuromation Labs Update
• Development Update
• NTK Update

Letter from the CEO

Artificial Intelligence (AI) is transforming how our world works. Andrew Ng, Chief Scientist at Baidu and Co-founder of Coursera, believes “Just as 100 years ago electricity transformed industry after industry, AI will now do the same.” By 2030, it is estimated that AI will add nearly $16 trillion dollars to the global economy. The impact of AI technologies is hard to fathom, and our world will be materially different in decades to come in unforeseeable ways.

Today, AI innovation is dominated by the major platform companies like Google, Facebook, Baidu, and others. They have built data moats and walls around their AI talent and in turn have dominated the AI ecosystem. These companies have proven that they do not respect our privacy and are focused solely on economic drivers. We believe AI should exist outside of walls and its true disruptive power should be leveraged to solve significant issues in areas like healthcare, food production, energy, and education.

A wave of disruption is now forming and the AI leaders of tomorrow are currently small, agile companies who are building AI in new and transformative ways. The core of the disruption is rooted in the emergence of a new class of applied AI developers who are rapidly acquiring the knowledge and skills to develop and apply AI to real-world applications. In the coming years, 10x the number of applied AI developers will emerge across the globe. The breadth and depth of talent in this network will surpass that of any platform company.

Here at Neuromation, our Mission is to “Transform industries by connecting, empowering, and growing the AI ecosystem.”

Our goal is to be the world’s destination for applied AI solutions, bringing together enterprise companies, applied AI developers, data providers and compute resources in an integrated and seamless platform. Our focus on a marketplace for AI and key enabling technologies, like synthetic data and distributed computing, will enable us to deliver solutions at a fraction of the time and cost.

By doing so, we hope to realize our Vision of “a world where AI is democratized and all people and businesses can contribute to and benefit from AI.”

We have built a company focused on developing real technology and products to impact the world in significant and positive ways. Our foundation is built upon a core set of values that govern our daily actions and drive our decisions. Our strategy is focused on three key themes 1) Cultivating an engaged and inclusive community 2) Creating a robust and efficient marketplace and 3) Building technologies and tools that make it easier and cheaper to build AI solutions

I came to Neuromation a few months ago after spending a decade in Silicon Valley building disruptive data-centric companies. I was incredibly impressed by the vision, passion, and capability of the team and I was honored to join as CEO. In my short time here, the team has been focused on building the foundation to help us succeed long-term. We started with putting in place a proper governance structure to ensure accountability to our contributors and shareholders. We have added two new Board members and put in place company policies.

Next, we recruited key senior team members adding leaders for the technology development and digital economy organizations. We will continue to hire senior folks and will look to add significant capability to our product and marketing functions. We continued to develop our core AI and synthetic data technology and our work has been recognized in Wired, TechCrunch and at leading AI conferences. We have also published research papers and built partnerships with a number of industry leading companies. Finally and importantly, we have sharpened our focus, as reflected in our new recently relaunched website, to delivering transformative solutions to our customers and empowering AI developers with the tools & technologies to make it easier and cheaper to build AI applications.

In this report we will provide a comprehensive set of updates. We are beginning a new phase of the business and will be increasing our communication to the community through various forms, including quarterly reports, regular blog posts, a new newsletter, and through our various chat communities.

I invite you all to reach out and engage our team. I look forward to the incredible road ahead and thank you for your tremendous support.

Best,

Yashar Behzadi

CEO, Neuromation

Introduction

Neuromation is excited to publish the first of our quarterly reports to update our community on progress to date. This report comes at a particularly exciting time in the development of the company, as Neuromation is now establishing itself as one of the leading companies in the field of AI, neural networks, and deep learning. In the first half of the year we completed many key milestones. We onboarded key new executives, including a new CEO and CTO. Neuromation was featured in such tech publications as Wired, TechCrunch, insideBIGDATA, Digital Journal, and Nasdaq. We substantially grew our team of world-class deep learning experts with true leaders in the field, adding 5 new data scientist/research engineer hires in Q1 and an additional 4 in Q2. We also recently relaunched our website, neuromation.io, and have refined and focused our custom AI solutions development offering to address broader enterprise demand.

In addition, development of the Neuromation Platform and associated toolsets continues at a brisk pace, as we will show in the Development Update below. The roadmap from the White Paper remains fundamentally in place, as our platform and offering continue to be focused on Neural Networks and Deep Learning Models and Frameworks, Synthetic Data and Distributed Computing. Certain key optimizations to our approach have been made in the past two quarters to better address identified customer needs. To address immediate enterprise deployment needs, we have developed an infrastructure-agnostic compute strategy that allows deep learning models to be deployed on various infrastructure environments. We feel these adjustments will improve our product offering, accelerate our time to market and increase the value of our offering to stakeholders. We look forward to sharing these exciting changes with you in the pages to follow.

We are also including an inaugural report on the performance of our NTK utility token, which we remain committed to supporting for use on all transactions to occur on the Neuromation Platform. We will continue to update the public on NTK dynamics and policies on a quarterly basis going forward.

Statement from Chief Research Officer, Sergey Nikolenko:

The current state of the art in artificial intelligence is defined by deep learning. The deep learning revolution occurred about 10 years ago, and by now, in most areas in machine learning, including computer vision, natural language processing, and many more, the best available solutions use various architectures of deep neural networks. Most exciting AI developments of late, including AlphaGo, self-driving cars, Google Translate, Facebook’s face recognition, and many others, are driven by deep learning.

Neuromation is distinguishing itself as one of the leading companies in this field. We already have 10 published papers with Neuromation affiliation to our name, including three papers on processing satellite imagery, five papers on medical imaging and a paper on molecular biology and drug discovery.

We have gathered a team of world class experts in deep learning, and are currently involved in projects that advance the state of the art in computer vision, natural language processing, and more. We have pioneered the use of digitally created data that mimics real-world sensory input, more commonly referred to as ‘Synthetic Data’ to power a variety of deep learning computer vision applications.,

In addition to this, we are also deeply involved in several other research projects related to computer vision. We successfully participated in the DeepGlobe challenge on processing satellite imagery and had 3 papers accepted for the DeepGlobe workshop at the leading computer vision conference CVPR. We are currently pursuing new exciting research projects in molecular biology and medical applications of deep learning.

Neuromation Labs Update

One of the most exciting components of the Neuromation story and what makes it unique among AI platform companies are its research labs, which are providing innovative AI solutions to clients and partners in a wide range of industries. This structure ensures that Neuromation is continually solving real world problems using the latest advances in the field, and also that the tools we are creating to assist in AI development are being used by leading AI practitioners on a daily basis, creating a positive feedback loop that we are confident will hasten the democratization of AI and spur creation of a community with the potential to change the world for the better.

During the first six months of the year, Neuromation continued to make progress in applied AI research through work on research partnerships signed in the first quarter, as well as major new research partnership agreements. Examples of this work follow.

Neuromation’s collaboration with OSA-HP on the use of synthetic data to accurately identify retail products (as seen in our whitepaper) was a success. We attained high levels of accuracy at product identification in a solution that can be trained and updated at a fraction of the time and cost of competing technologies. OSA-HP, a retail industry consortium of multinational FMCG and food and beverage companies, is currently using Neuromation’s demo worldwide for all of their presentations and expos and it is anticipated to be implemented throughout their clientele after completion.

Neuromation is also focusing on AI applications for manufacturing, traveling to Shenzhen China to partner with Linfinity Foundation to create AI applications for manufacturing quality control and supply-chain management.

In the health sciences arena, Neuromation is conducting research in cooperation with EMBL — the European Molecular Biology Lab, on the use of synthetic data for the automated analysis of medical imagery. According to Stratistics MRC, the global medical imaging market is currently over $30bn and expected to reach $45.1bn by 2022. Neuromation is proud to be at the forefront of AI research in this exciting and important area.

In a related field, Neuromation announced that it is conducting research in coordination with Insilico Medicine Inc. on automated drug discovery and design with machine learning to generate and modify molecules. This is also a dynamic and growing industry, currently valued at almost $40bn annually.

Another research partnership announced during the quarter was a project with Monbaby on developing a computer vision system trained on synthetic data for pose estimation of babies while sleeping.

On the robotics front, Neuromation has partnered with TRA Robotics on training artificial-intelligence powered robots with synthetic data. Results to date have been extremely encouraging with robots being successfully deployed in real world environments and performing with a high degree of accuracy.

Let’sEnhance seeks to dramatically improve the capability for upscaling the resolution on old, low-res or blurry photos. Through its partnership with Neuromation, Let’s Enhance employs Deep Convolutional Neural Networks trained on a dataset of real and synthetic images, allowing them to learn typical features of objects so the application can add extra details that were not present in the original. This moves beyond what was possible with the previous state of the art — bicubic interpolation — which is employed by industry leading platforms like Adobe’s Photoshop.

Additional areas of research of Neuromation Labs can be seen in the list of published papers that have Neuromation researcher affiliation. This list currently includes papers on satellite imagery analysis, medical diagnosis and robotics, with more on the way.

Papers with Neuromation affiliation

1. Medical Concept Normalization in Social Media Posts with Recurrent Neural Networks, E. Tutubalina, Z. Miftakhutdinov, S.I. Nikolenko, V. Malykh, Journal of Biomedical Informatics, vol. 84, 2018, pp. 93–102
2. Land Cover Classification With Superpixels and Jaccard Index Post-Optimization, A. Davydow, S.I. Nikolenko, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, (CVPR 2018 Workshops), 2018, pp. 280–284
3. Building Detection from Satellite Imagery Using a Composite Loss Function, S. Golovanov, R. Kurbanov, A. Artamonov, A. Davydow, S.I. Nikolenko, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, (CVPR 2018 Workshops), 2018, pp. 229–232
4. Land Cover Classification from Satellite Imagery With U-Net and Lovász-Softmax Loss, A. Rakhlin, A. Davydow, S.I. Nikolenko, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops, (CVPR 2018 Workshops), 2018, pp. 262–266
5. Diabetic Retinopathy detection through integration of Deep Learning classification framework, A. Rakhlin, bioRxiv, June, 2018, DOI: 10.1101/225508, https://www.biorxiv.org/content/early/2018/06/19/225508
6. Deep Convolutional Neural Networks for Breast Cancer Histology Image Analysis, A. Rakhlin, A. Shvets, V. Iglovikov, A.A. Kalinin, Proceedings of the 15th International Conference on Image Analysis and Recognition (ICIAR 2018), DOI: 10.1101/259911, https://www.biorxiv.org/content/early/2018/04/02/259911
7. 3D Molecular Representations Based on the Wave Transform for Convolutional Neural Networks, D. Kuzminykh, D. Polykovskiy, A. Kadurin, A. Zhebrak, I. Baskov, S.I. Nikolenko, R. Shayakhmetov, A. Zhavoronkov, Molecular Pharmaceutics, 2018, DOI: 10.1021/acs.molpharmaceut.7b01134, https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.7b01134
8. Angiodysplasia Detection and Localization Using Deep Convolutional Neural Networks, A. Shvets, V. Iglovikov, A. Rakhlin, A.A. Kalinin, bioRxiv, DOI: 10.1101/306159, https://www.biorxiv.org/content/early/2018/04/23/306159
9. Automatic Instrument Segmentation in Robot-Assisted Surgery Using Deep Learning, A. Shvets, A. Rakhlin, A.A. Kalinin, V. Iglovikov, bioRxiv, DOI: 10.1101/275867, https://www.biorxiv.org/content/early/2018/03/03/275867
10. Pediatric Bone Age Assessment Using Deep Convolutional Neural Networks, V. Iglovikov, A. Rakhlin, A.A. Kalinin, A. Shvets, bioRxiv, DOI: 10.1101/234120, https://www.biorxiv.org/content/early/2018/06/20/234120

The activities of Neuromation Labs to date, including the research partnerships and the topics and industries involved, represent massive new potential market opportunities for Neuromation. They are crucially important in terms of proven capabilities gained, processes developed, and a solid record of success. They have also contributed to Neuromation’s ability to attract a world class group of AI researchers as the best minds in the field are attracted to the opportunity to work on a wide array of large real world problems.

This early work has also been key in assisting in development of our toolsets for simplifying and automating key tasks involved in AI research and development.

Expanded Availability of Neuromation Custom Solutions

In response to enterprise demand and interest in Neuromation’s AI capabilities and platform, Neuromation is now expanding its custom solutions development services to include paid customer engagements for a broader range of corporate clients. These engagements will serve as prototypes for future marketplace engagements on the Neuromation Platform and jumpstart community involvement. In support of this effort, our redesigned website, neuromation.io, both addresses the needs of enterprise customers and solidifies our thought-leadership in synthetic data and deep learning.

Building upon the experience, track record, capabilities and toolsets developed in Neuromation Labs during the past half year, Neuromation now has the processes, tools and team in place to accurately assess time and cost factors and likelihood of success necessary to truly operate at scale in the commercial sector.

Utilizing its team of in-house research engineers and software engineers, Neuromation is able to provide crucial insights into how companies in a range of industries can utilize artificial intelligence and to create custom AI solutions using synthetic data.

Interest in these services has been overwhelming, with dozens of expressions of interest received throughout the quarter. Neuromation is currently working its way through these prospects by prioritizing them according to clarity of vision and size of prospective engagement.

Although all customer engagements are different, a typical client engagement undertaken by Neuromation or other marketplace participants may include strategy formulation (1 month), R&D (2 months) and product development (4 months). A longer engagement may involve a complete process of reimagination.

We will endeavor to provide insight into the progress of this work in future quarters, as metrics stabilize following launch.

Each client engagement has the potential to provide key experience in new industries, potentially leading to ground breaking R&D, new product opportunities, and always informing creation of our toolsets and platform. Clients in leading industries will also serve to seed the broader ecosystem to catalyze the growth of Neuromation’s AI marketplace.

Finally, in order to ensure that Neuromation’s interests remain aligned with holders of its NTK utility token, Neuromation confirms its pledge that during the current year 100% of payment transactions related to corporate AI development will occur in the form of NTK purchases.

Select New Hires in Deep Learning Team:

Rauf Kurbanov, Lead Researcher

Rauf graduated from the St. Petersburg State University with a BS in Software Engineering in 2015, followed by a Masters in Machine Learning at the St. Petersburg Academic University in 2017. Rauf developed production-ready machine learning models for NLP & Speech Recognition at Behavox. Then, for two years, Rauf served as a full-time researcher at Jetbrains Research as a member of the Machine Learning and Applications research group. Rauf is a lecturer on his own “Deep Learning” and “Speech Generation & Synthesis” courses. He is completing on his Ph.D. thesis on generative models & synthetic data at SPbAU RAS.

Aleksey Artamonov, Lead Researcher

Aleksey graduated from the Astronomical Department of the Faculty of Mathematics and Mechanics of the St. Petersburg State University. He also graduated from the Computer Science Center, where he received an education in computer science, data science and computer vision. Aleksey has worked as an engineer at Yandex for 4 years, where, among other things, he wrote distributed systems in Java. He is a lecturer on computer vision at CSC and SPbAU RAS. Aleksey joined Neuromation in order to focus on the research and development of deep learning and computer vision.

Elena Tutubalina, Researcher

Elena graduated with a Specialist degree with Distinction (Master of Science equivalent) in Computer Science from Kazan Federal University, Institute of Computational Mathematics and Information Technologies (2012). She completed her Ph.D. thesis in Computer Science at Kazan Federal University (2016). Her particular area of expertise is natural language processing and opinion mining. Elena is the author of 30 research papers, including papers on top NLP conferences such as ECIR and in top academic journals (Q1 Web of Science). In 2015, she was one of the organizers of the shared task on sentiment analysis in Russian, the SentiRuEval 2015 competition. In 2017–2018, she delivered a course on natural language processing for undergraduate students at the Kazan Federal University. She has experience in managing and participating in research projects supported by the Russian Science Foundation and Russian Foundation for Basic Research.

Alexander Rakhlin, Researcher

Alexander graduated with a Masters degree in Computer Science from the National Research University of Electronic Technology (MIET), Department of Microelectronics and Technical Cybernetics in 1994. He began his career in fintech, specializing on research and algorithmic trading. In 2013, he completed CaltechX course “Learning from Data”, with honors, and fell in love with Machine Learning. He has participated in various Data Science contests and commercial projects, many of them directly connected with medicine and biology, especially medical imaging. Alexander earned the distinctive status of Kaggle Master. He is the author of research papers on medical applications of Deep Learning in ophthalmology, radiology, microscopy imaging, and he is a teaching assistant at Coursera.

Development Update

In late March, 2018, Neuromation concluded an extensive executive search process for a new CTO, resulting in the hiring of Artyom Astafurov based in New York City.

Artyom has over 15 years of executive engineering experience with major technology companies and was previously the Managing Partner in charge of the Distributed Systems practice at DataArt, a global technology consultancy and solutions developer with over 2,500 employees. Most recently, he served as VP of Engineering at Opentrons Labworks, a robotics company manufacturing liquid-handling robots for the biomedical industry.

The first task of Neuromation’s new CTO was to conduct a top to bottom review of all development activities at Neuromation and to begin making key optimizations. Activities included:

• Full analysis of codebase to date
• Site visits to all offices, interviews with team leads and all developers
• Streamlining of product engineering lifecycle
• Fine tuning of team composition and overseeing of all new development hires

Furthermore, under new technical leadership and with a streamlined team, in April 2018 Neuromation began a systematic evaluation of the proposed technologies preliminarily described in the Neuromation Whitepaper, including:

• Distributed compute based on crypto mining hardware
• Portability of synthetic data approach across multiple use cases
• Surveying state of the art research in generative models to produce synthetic data
• Toolsets and underlying technologies required for development and training of machine learning models on a shared platform environment that will enable future marketplace operations

Approaching the issue of distributed compute using crypto mining hardware as a hypothesis requiring detailed verification and confirmation, Neuromation’s development teams ran a series of deep learning benchmarks comparing various configurations of crypto mining rigs and infrastructure offered by public cloud providers such as Amazon Web Services (AWS) and Google Cloud. These results showed a 2x cost and efficiency improvement for the crypto environments. When compared against Google Cloud, costs were almost the same as the crypto miner distributed compute set up.

Neuromation also conducted testing of competing cloud platforms to gauge their suitability for AI research, creation and collaboration, collecting extensive data and insights on best practices for utilizing these resources.

Finally, Neuromation consulted with a wide range of AI researchers and developers as well as enterprises actively interested in deploying AI solutions.

As a result of this process, it was decided that to meet both enterprise and AI developer needs and to ensure compatibility with a broad set of compute infrastructure, the best course for the platform development was to utilize stable managed platform-as-a-service environments allowing us to boost development velocity and shift focus to core functionality and developer experience. This approach will also allow keeping the platform portable to other infrastructures including mining rigs in the future.

Developer Experience and Community Building

The main goal of Neuromation’s development efforts to date has been to jumpstart community growth by focusing on developer experience — with the aim of bringing value to market early on and to secure early adopters by solving their problems.

By “developer experience”, we mean making it dramatically easier, cheaper and faster to develop AI solutions.

Element AI, an independent lab in Canada, recently estimated that there are 22,000 serious AI researchers worldwide. While this figure is growing rapidly, it is still roughly one tenth of one percent of the number of software engineers worldwide.

By providing tool sets that allow a broader range of skilled software developers to meaningfully participate in AI development, Neuromation believes it can multiply the amount of human talent and creativity being applied to AI problems by 1000x and increase Neuromation’s total addressable market by an order of magnitude. This is what we mean by the power of democratizing AI.

Key problem areas that Neuromation aims to simplify:

• Rapid evaluation and development of deep learning models
• Training and deployment of deep learning models
• Infrastructure management
• Collaboration
• Synthetic data creation
• Tracking of performance metrics
• Dataset storage management
• Batch and stream inference

Tools already developed by Neuromation to date include services to store and retrieve training data sets, to run and monitor training and inference jobs, to simplify management and orchestration of GPU resources. These tools are also being used on a daily basis by Neuromation research engineers.

Additional development priorities for Q3 include creation of visual tools for simplifying repetitive tasks such as dashboarding, administrative tasks, permission management and an improved AI-domain-specific command line interface for developers.

This suite of tools will be designed and integrated by Neuromation for the specific purpose of AI development, and will be available for all users on the Neuromation Platform. The Neuromation Platform is agnostic to frameworks — models can be built using Pytorch, Tensorflow and other frameworks and libraries. Tool Sets will be portable between public clouds, allowing compute resources from third-party providers such as Amazon, Google and others to be integrated seamlessly. This will allow users to take advantage of subsequent performance improvements and price reductions as they occur and to choose the best resources for a specific job, use case, or geography. Trained models that move into production will run directly on any suitable infrastructure.

Our cloud agnostic approach will also allow the option for customers to use private cloud if necessary. This is a key requirement to meet the needs of certain large enterprise customers. Among enterprises that use private cloud, one of the main reasons given is quantity of data — in certain cases, there is simply too much data to move to the public cloud easily. Data security is another main reason. Neuromation’s approach could allow for AI development to occur wherever a company’s data resides, and on its own hardware, which are both important advantages.

NTK Market Update

Neuromation remains committed to utilizing its ERC20 utility token NTK as the exclusive means of conducting transactions on the Neuromation Platform and maintains its commitment to always consider the value to its token holders first when considering any new product or service offering in the future. To this effect, 100% of revenue from current corporate AI development projects will be used to repurchase NTK, reducing the total outstanding amount of tokens on the market.

In the months following the NTK token sale, Neuromation made significant efforts to ensure that NTK has appropriate liquidity by listing on several reputable cryptocurrency exchanges, with operations focused on North America, Europe and Asia.

Current exchange listings:

• Tidex: Listed February 20
• HitBTC: Hong Kong based exchange, listed May 7
• BCEX: China focused exchange, listed May 3

Community efforts have resulted in additional listing on:

• Yobit
• IDEX
• Cobinhood

Against the backdrop of significant losses to overall cryptocurrency market capitalization in the first two quarters of 2018, NTK has similarly faced negative pricing pressure. In all, NTK dropped by 76% from opening of trading on February 20 to June 30, 2018. In Q1, NTK traded down 54%, vs. -50% for BTC, -47% for ETH and -78% for XRP. In Q2, NTK continued to trade poorly, losing 47%, vs. -8.4% for BTC, while ETH recovered 14.6%.

General cryptocurrency pricing catalysts for the first half of the year are fairly well understood, and include steadily increasing regulations in Asia, including new regulations in Japan, Singapore, China, and South Korea. New restrictions on cryptocurrency advertising announced during the quarter on the major search and social networks, such as Facebook, Google and Twitter, and may have also had a negative impact.

In terms of NTK-specific pricing pressure, the addition of two Asia-focused exchanges in Q2 may have increased NTK’s exposure to negative Asia-related cryptocurrency news, but it is also possible that the continued negative price pressure was due to short term cryptocurrency investors who had purchased NTK hoping for a quick gain abandoning the Neuromation story in the absence of visible short term catalysts.

A key concept to keep in mind as we look at NTK pricing is that NTK is a utility token created for use on the Neuromation Platform. It is not a security token, nor a cryptocurrency. As the Neuromation Platform is still in development, it is logical to assume that current demand for the NTK is dominated by speculation.

Furthermore, if we look at the cohort of the top 25 largest ICO’s of 2018 for which pricing is available on Coinmarketcap (see table below), average YTD price change for the group was -65%, suggesting that NTK performance was not an outlier and showing a remarkably similar trading pattern among utility tokens released this year.

Largest ICO’s of 2018
Ordered by performance

Looking to the future, potential positive catalysts for NTK may include public release of the Platform, feature updates, or positive user growth metrics. Neuromation will make every effort to keep our community informed of these and other relevant pieces of news and information regarding the company and its products. We will also endeavour to provide expectations regarding release dates or user growth targets as these events become closer and more predictable.

Corporate Policies

We would like to use the opportunity presented by this update to restate our long-term commitment to defending the interests of the NTK token holder community.

Neuromation believes that global token markets will be increasingly regulated, and that participants who do not prepare for new regulation will face existential risks to their business models and token economies, as well as a destruction of the interests of their token holders. Neuromation believes increasing clarity of regulation will be a positive driver for our company, and other market participants who are dedicated to long term development of their businesses and communities.

For this reason, Neuromation has chosen the path of pro-active adherence to potential future regulations, and we are dedicated to two major principles in this regard:

• First is the principle of self-regulation in expectation of the implementation of regulatory frameworks around the world
• Second is the principle that the interests of Neuromation, its owners and employees are aligned with the interests of NTK holders

Neuromation has / is taking the following actions in order to adhere to these principles:

First, Neuromation has implemented internal corporate policies regarding the activities of owners, employees and advisors with the NTK. These policies are in line with global standards, prohibiting employees and advisors from engaging in trading of the NTK based on insider or confidential information.

Furthermore, NTK held by founders, employees and advisors have agreed to be subject to lock-up, i.e. the prohibition of selling NTK in the market, until at least January of 2019.

Second, Neuromation engaged Baker & McKenzie, who produced a Legal Opinion confirming the status of NTK as a utility token. This is an important step towards Neuromation’s goal to join the largest and most highly regulated crypto exchanges, market infrastructure providers who can provide sufficient liquidity for NTK to allow our users to confidently and seamlessly transact on the Neuromation Platform.

Third, for the coming year, Neuromation commits that 100% of revenue from client engagements for custom artificial intelligence solutions will be executed in the form of NTK purchases, fully aligning our corporate development work with our Platform and the interests of our NTK holder community.

Finally, we would like to update the community on our burn policy. As described in the whitepaper, the burn policy calls for a percentage of all NTK transactions on the Neuromation Platform to be removed from circulation (burned). The total burn, therefore, will be dependent on the size and frequency of NTK transactions between counterparties on the Platform. It should be restated that only NTK held by Neuromation will be burned. At no time will the NTK of any other holder be removed from circulation by Neuromation.