GraphGrail Rating Review


Investment Rating

Expire date : 25.08.2018

We assign the GraphGrail AI project a "Stable" rating.

Graphgrail AI is a decentralized platform that enables the design of applications based on artificial intelligence and blockchain technology without programming skills. On the platform, placing tasks and receiving orders for data processing will be possible; solutions and data may be sold on the built-in marketplace.

Work on the project has been underway since 2014. The development is being carried out by a little known team from Russia; however, given the number of existing developments presented to the general public in the MVP, there are no doubts about the team's skills.

GAI tokens will be used as the internal currency of the platform. The tokenomy mechanisms proposed by the team will help to increase the value of the token in the long term. A buyback mechanism that is planned but not disclosed in official documentation could provide additional token support. The amount of funds that the team will be able to use for buyback is unclear, as there are no financial calculations.

Risks for the project lie with its technological and marketing aspects. These risks will be detailed in the relevant chapter of this review.

Graphgrail AI is a decentralized platform enabling the design of applications based on artificial intelligence and blockchain technology without the need for programming skills. Posting tasks and receiving orders for data processing will be possible on the platform; solutions and data may be sold on the built-in marketplace.

The GraphGrail AI team has its own technological development in the field of artificial intelligence for working with large arrays of text data. The team is using a development method based on the biological approach offering neural networks to gather data and model texts. The team is using best practices of the development industry including various architectures of neural networks (based on the biological approach), syntactic and semantic data processing, thus, allowing for the platform to gather data and build topic and semantic text models.

Aspects of the technology proposed by Graphgrail AI have already been created, tested and successfully used by the project team for business and public administration tasks. The site offers an MVP, which enables understanding the possibilities of the technology in its first approximation.

For the Graphgrail AI project, blockchain performs not only a function for choosing the kind of investment attracted, but also provides the project with an internal currency. Project tokens will also store results for users received when training a neural network, loading, and/or distributing data.

The utility token satisfies SEC conditions. Legal development for the project was carried out by Juscutum Attorneys Association, and it has also worked out the legal risks of the project. The legal shell company for the project has been in operation for several years, which also contributes to reducing the risks of investing in Graphgrail AI. The jurisdiction is the British Virgin Islands. The majority of the developers are based in Rostov-on-Don, Russia.



Token: GAI (ERC-20)

Platform:  Ethereum

Volume of placement: 270,000,000 GAI

Token distribution: founders — 22%, for sale — 50%, bonus fund — 25%, partners — 1%, bounty — 2%.

Round 1: Pre ICO (closed on 20/07/17)

  • Cap: $7000

Round 2: Pre ICO (closed on 16/10/17)

  • Price: $0.02
  • Cap: $200,000
  • Bonuses: 15–25%%
  • Minimum Buying Transaction: $10,000
  • Maximum Buying Transaction: $100,000

Round 3: Public sale

  • Volume of placement: 120,000,000
  • Start: 19/02/18
  • End: 15/04/18
  • Soft Cap: $2m
  • Hard Cap: $12m
  • Price: $0.1
  • Bonuses: 15–35%%

Raised on: $540,000

Funds allocation: 45% development, 30% marketing, 5% legal, 20% Ai Lab.


  • Passing KYC and whitelist registration are necessary for participation.
  • Investment funds: Reliable data are absent; a well-known venture investor, Alexander Borodich, is a co-founder.
  • Bounty campaign.

In this section, we usually discuss the existing and planned for implementation pool of the project services and focus on the technical issues.

Variants and examples of the possible application of neural networks and machine learning for business tasks are very broad:

  • Forecasting; risk assessment. (Forecasting demand, volume of sales, average check, frequency of sales, loading of equipment for the optimization of cash quantities, storage places and other resources).
  • Search for trends and correlations. Forecasting further development of a system and predicting possible changes.
  • Recognition of photos, videos, audio content. Various services and online applications with the use of recognition technology. (Example of the "LiarScan" project for lie detection).
  • Machine learning for computer system dialogues. For automation of activity in online chats, as well as for telephone operators and instant messengers. Development of chat-bots.
  • Finding and analyzing  facts and events in external sources will help Oracles to provide smart contracts with data about changes in the real world. It is a fundamental solution facilitating transition of businesses from traditional contracts to smart contracts.

The GraphGrail AI startup aims to provide easy access to the above features for various business entities that lack relevant expertise in IT and programming.

When discussing investment in the GraphGrail AI project, it is necessary to understand that at present a considerable portion of the technological work has already been carried out, the blockchain mechanism for big-data storage has been developed, and there is a working model of the analyzer using neural networks (artificial intelligence). Moreover, the project has successful experience of monetization of its technology working with several large companies and governmental bodies.

GraphGrail AI will provide a simple interface for creating an application model and subsequent machine learning.

On the GraphGrail AI platform one will be able to create and train neural networks using a user constructor. It is expected that business executives, startup owners, developers, data experts and many others will be able to create their own applications for integration with their own services and applications. The second possibility for the platform is a full cycle of work with big data, from gathering and marking up to final result.

Currently, the main task for the team is to combine these services into a single ecosystem, create a site, marketplace and mobile application.

The GraphGrail AI project will offer users four key services:

  • GraphGrail AI designer
  • GraphGrail AI labellance
  • GraphGrail AI marketplace
  • GraphGrail AI Lab.

Graphgrail AI Designer is a user builder for creating applications. The Designer can create and train neural networks for various tasks including complex classification, using Google TensorFlow and other tools. For business this means simple development of chat bots, analytical products, products and services in media, determining the authorship by style of the text, exact identification of emotions from statements. Moreover, the designer provides an opportunity for specialists lacking in programming knowledge to work with the platform.

Graphgrail AI Labellance — an interface for data markup. Users will be able to mark up arrays of text data in their language and extract hidden knowledge that facilitates management decisions. Graphgrail AI Labellance will also enable one to create markups to order.

The Graphgrail AI Marketplace is a marketplace for language models with a possibility for monetization and payment for requests.  The marketplace will enable users to buy and sell ready marked up datasets for training neural networks.

Graphgrail AI Lab — in this laboratory for deep machine learning, artificial intelligence researchers and experts in the analysis of data from around the world will be able to develop and test new and promising solutions.

In addition to the above services, Graphgrail Ai will also offer users supporting services such as:

  • An automated smart contract execution system, operating through a cross-blockchain ecosystem, webAPI and external data sources.
  • The implementation of ready-to-use sets of semantic categories (category-subcategory, taxonomy, part-whole).
  • Implementation of blockchain for the quality control of data markup (proof-of-quality-work).

Answering the key question of this paragraph — "Does the project need blockchain?", we emphasise that blockchain acts as technological support for the ecosystem and of course ensures its integrity.

The funds raised as a result of the initial offering will be spent on the completion of the product development process. Among other things, this includes the full launch of the platform with API access, launch and testing of the language models marketplace as well as the support of prospective developers creating applications on the startup platform.

Market analysis

Artificial intelligence is technology that is intended for the study and development of software for intelligent machines. Artificial intelligence technologies are widely used in various industries. Demand for solutions involving artificial intelligence is growing due to the need for companies to increase productivity. This factor will play a key role in the development of this market in the coming years.

The world market for artificial intelligence is segmented by solution type: Electronic computing systems, artificial neural networks, automated robotic systems, embedded systems and digital assistants.

The number of projects related to artificial intelligence and machine learning has grown globally several times in the last two years. In 2015, large companies reported the existence of 17 such projects, in 2016 another 71 projects were launched, in the first half of 2017 — 74 projects. Thus, according to the results of 2015-2017, the total number of initiatives has reached 162. 28 countries and 20 industries are involved in their implementation.

85% of these projects have already been implemented, another 15% are in the planning stages or pilot phase, and 60% of initiatives are in the public sector at this stage. In 85% of cases, projects are carried out to order for a large business.

The United States is the leader in the number of implementations of artificial intelligence and machine learning technologies. Second is the UK, which applies these technologies in large investment banks, and India which uses them in work for foreign customers.

International Data Corporation (IDC) estimates that the volume of the global market for cognitive systems and artificial intelligence technologies in 2016 amounted to approximately 7.9 billion USD. In 2017, the market reached a volume of 12.5 billion USD, which corresponds to an increase of 59.3% compared to 2016.

IDC analysts believe that the average annual growth rate in complex interest rates (CAGR) by the end of 2020 will be at the level of 54.4%. As a result, in 2020 the volume of the industry will exceed 46 billion USD.

Currently, artificial intelligence is the most important field of IT research. Electronic intelligence, in particular, will help to analyze the huge amount of data that will be generated by IoT devices. Experts estimate that by 2020 more than 50 billion machines and devices capable of connecting to the network and exchanging information among themselves will be operational worldwide.

In 2017, 1.74 billion USD and 1.72 billion USD were the share of the trade and banking industries respectively. Researchers have spent more than a billion dollars on artificial intelligence in discrete vs. continuous production and health care. At the same time, trading companies not only invest more funds, but also increase their investments more quickly. The average growth in this segment was 58.8% per year.

The most popular areas of artificial intelligence and cognitive systems are the creation of automated customer service agents ($1.5 billion will be allocated to this) and diagnostic and repair systems ($1.1 billion). The fastest growing segments of customer recommendation systems (96.6% per year), public safety and emergency response (96.2%) and intelligent process automation (69.9%).

It should be noted that about half of these investments are in software, about a third are in services, and the equipment segment is only 18.8%.

Key trends in the artificial intelligence market:

  • Democratization of instruments will give access to artificial intelligence to more companies. A recent Forrester study among organizations and professionals in the technology field has shown that 58% of them are exploring the possibilities of artificial intelligence, but only 12% use these systems. This is partly because they are starting to be used only now, and because the technology is in the early stages of development and is not easy to use. Working with these systems requires a set of specific skills and a specific approach.
  • The emergence of a large number of general-purpose systems.
  • The economic impact of increased automation will be a topic for discussion.
  • Further complication of systems that prevent excess information.
  • Increased focus on ethics and privacy.

Thus, Graphgrail Ai Lab operates in a market with a volume of 12.5 billion USD and a projected growth rate of 54.4% over the next 3 years. The Graphgrail AI Lab project is aimed at solving one of the key problems for the industry — that of democratization.


According to a study by Transparency Market Research the leading players in the market of artificial intelligence solutions are IBM, Intelliresponse Systems, Nuance Communications, EGain, MicroStrategy, Brighterion, Google, Microsoft, Next IT and QlikTech International. Most projects in the field of artificial intelligence are extremely complex and expensive for most users.

The idea of the democratization of artificial intelligence tries to interpret many projects in its own way. Frameworks like Facebook and Howdy Slack are trying to become a kind of Visual Basic artificial intelligence, promising the simple development of intelligent conversational interfaces without the requirement of a high degree of developer training. Tools like Bonsai, Keras and TensorFlow simplify the introduction of deep learning models. Cloud platforms, such as the Google and Microsoft Azure interfaces, enable one to build intelligent applications without having to worry about configuring and maintaining an appropriate infrastructure.

Nevertheless, projects based on open decentralized platforms and blockchain technology are now coming to the foreground. A comparison of such projects with classic platforms is presented in the table.


Open platform

MS Azure

IBM Watson

Yandex Toloka

Dandelion API

Working without programming skills






Ready-to-use sets of semantic categories


Payment required

Payment required

Payment required

Payment required

Automation of a typical business work-flow


Salaried developer required

Salaried developer required

Salaried developer required

Salaried developer required

Ease of change/customization of a solution for oneself


It is necessary to order a special solution

It is necessary to order a special solution

It is necessary to order a special solution

It is necessary to order a special solution


Currently development of several projects similar to Graphgrail AI based on decentralized platforms is also underway.

Opensource (Gluon) — an interface for creating machine learning models using pre-assembled and optimized components, building blocks that can be used together with Amazon and Microsoft platforms. Ideally, it should facilitate the process of developing models for beginners and accelerate the creation of complex systems for experienced professionals. Gluon is now compatible with Apache MXNet, an open-source deep learning platform, and Microsoft is committed to its compatibility with its Cognitive Toolkit tool. is a platform that enables creating an artificial learning environment for neural network deep learning. These models are then used to train and improve algorithms. The idea of Neuromation is to create a platform for the practical use of its own scientific developments in the field of design of neural networks and artificial intelligence systems. The main business of the platform will be associated with compiling classified data sets for the training of neural networks. Typically, data with manual object tagging is used to train neural networks, but obtaining such data is very costly. Neuromation offers to replace real data sets with synthetic data, which is quite suitable for training neural networks in certain areas of business. To generate synthetic data, it is planned to use the computational capacities of existing cryptocurrency mining farms based on graphic video cards.

dBrain is a decentralized blockchain platform for crowdsource data generation for training AI-based solutions, based on neural networks. The platform carries out dataset markup. When data is marked up, the platform finds a developer through open competition, who creates a neural network algorithm according to the technical specification of the customer. The developer receives a fixed payment (from 1 thousand to 300 thousand dollars. for a private network — or a percentage of the cost of the turnover, if the network is public. dBrain checks the finished solution and the business connects to it through the API.

NeuroSeed is a unified ecosystem for the sale of machine learning models. Each user can be sure of receiving paid data or payment for their intellectual property. As a result, data exchange markets and ML-solutions are created around the platform; computing power and various data storage methods are provided as well as a data market.

Graphgrail AI therefore operates in a highly competitive market characterized by the presence of a large number of startups at an early stage of development, which could compete seriously with existing market leaders.

The key difference of the GraphGrail AI platform from the decentralized projects listed above lies within the following two factors:

1.  The focus on data processing, which potentially allows to provide solutions of better quality compared to those that will soon appear on other platforms which in trying to cover all the range of AI solutions (photo, video, time series mining) lose in quality. GraphGrail AI is developing in consecutive stages, and after the AI ecosystem is proven to work on texts, the project scaling and expansion to other market segment would be possible.

2.  Own AI solutions and a builder for them allowing for the platform to provide businesses a full cycle of work with data, and this often distinguishes the platform from competitors usually offering 1 or 2 solutions that do not allow businesses to receive a valuable end product. For example, platforms that cannot build their own solutions and only have a marketplace for selling them will face the difficulties with attracting users to an empty platform. Also their is an issue of integration , that is, even if a business client buys an ML-solution, they still have to integrate it into their system because otherwise its value significantly decreases.

GraphGrail AI provides a single solution for analyzing text data. The Graphgrail AI team consists of 30 specialists in data-science, natural language processing, programming, marketing and other fields. The founder is Victor Nosko. The position of key advisor and CMO is occupied by a venture investor, Alexander Borodich.

Key team members:

Victor Nosko — CEO and founder. Python Developer, Django framework. Data-science specialist, NLP stack: NLTK + Celery + Pymorphy2 + GLRparser, etc. Victor has over 6 years of experience in development and deep learning, and is experienced in Google TensorFlow.

Alexander Borodich — Venture investor, CMO. Futurist, business angel, founder of VentureClub, MyWishBoard, MyDreamBoard, and SuperFolder. Partner at Future Action, founder of, and Universa. Universa held its ICO in 2017; it was subjected to an active information attack online, which damaged the reputation of Mr. Borodich in the crypto community.

Anton Smetanin — Fullstack web developer. He is responsible for backend development. He has more than 7 years’ experience in this field.  Main languages and frameworks used: PHP (Yii), Python (Django), Javascript.

Alexander Gusarin is a Python and Data Science developer. Responsible for the development of machine learning programs and Python programming.

Zakhar Ponimash — Consultant on neural networks. He works with neural networks and artificial intelligence. Game developer, based on the XNA framework, TCP/IP chat, bot chat, text comprehension systems.

Semyon Lipkin — Developer of Python and Data Science. Sphere of activity — development of algorithms of machine learning using the Python language.

Maria Tarasova — Journalist. Candidate of philosophical sciences, specializing in simulation, data mining and statistical data analysis. She was awarded a scholarship by the President of the Russian Federation and the Government of the Russian Federation for major contributions to science; she is the author of more than 60 research works on the modeling of social processes, an active participant in 5 grants from the Russian fund Fundamental Research, participant in more than 10 national and international conferences.

Marina Parinova — HR manager. Responsible for IT recruitment.

Nikita Buyevich — Frontend Developer. He is responsible for creating user interfaces.

Once again we emphasise that the established team has successfully completed a number of projects for business and government. Most of the team has solid experience in the implementation of scientific and business projects. The team has all the necessary relevant experience to implement its project. However, we identify marketing and finance as potential shortcomings.

Graphgrail AI is selling GAI tokens during the ICO. GAI is a utility token that acts as the internal currency for the system.

To be able to access the system, users (primarily business users) will have to purchase a certain amount of GAI tokens — from 5 to 10,000. These tokens can then be spent by the user on internal services of the platform — collection, cleaning, data marking, custom settings for training a neural network, etc.

Users who receive tokens as payment for their services will be able to convert them to fiat or other cryptocurrency. However, due to an obligation for users to make a one-time purchase of a relatively large volume of tokens, the project will likely achieve a permanent excess of demand for the token over its supply, provided that the product is successfully implemented.

The token has no other functionality.

By and large, the team could replace the GAI token with any liquid cryptocurrency and use it as the internal currency. In other words, GAI tokens should primarily be considered as a mechanism for funding the project.

We have already noted that the proposed mechanism obliging users of the platform to buy a certain amount of tokens at a time from the market to gain access to the platform's functionality, will help to permanently ensure demand for GAI exceeds supply. This is only given the emergence of a steadily growing utility demand.

In this regard, optimism is engendered by the fact that work on the project has been underway since 2014 according to the roadmap; the team has already achieved certain developments, and new services will be introduced with enviable regularity. Key elements of the platform will start functioning before October 2018, which should ensure the relatively short-term appearance of infrastructural demand for the token.

The project team has shared with us its plans to buy back tokens from the market not more often than once a quarter. Bought back tokens can be either burnt or used for platform purposes.

The burning of tokens could have a positive impact on their price, whereas a return to circulation is unlikely. Moreover, GraphGrail AI has a reserve fund consisting of 25% of the initial issue. The plans of its use as yet exist only as a first approximation — it is suggested that it will be used for attracting users and developers to the platform as additional motivation and for accumulation of datasets, libraries, algorithms, i.e. intangible assets for the benefit of the team. The team reported that 10% of the reserve fund will be allocated to the platform needs in attracting independent developers, another 10% will be spent on funding and holding competitions on improving algorithms, and the remaining 5% will be spent on airdrop among the partner companies holding the token.

We have described the risks for the token in the previous chapter. Below we will concentrate on the risks of the project itself and its ICO. 

In such projects there are always risks of a technological nature, i.e. risks of technical realization. In this case, the team has long been working on the project which increases the likelihood of success. However, it remains unclear whether the innovations being developed will be applicable in practice in the near future.

Another project risk is the fact that the team has not involved anyone prominent in the crypto community or business environment. An exception is Mr. Alexander Borodich, whose previous project, Universa, faced active criticism, albeit ambiguous in nature. However, GraphGrail AI’s hard cap is also small compared to other ICOs, so this risk also should not be considered significant.

Marketing and promotion of the ICO are also among the risks of the project. At this stage the traditional metrics for estimating activity of the ICO campaign (Telegram, Bitcointalk, etc.) are at an extremely low level. Publications in the press about the project are also few and far between. The project is niche and specialized. In this sense, there is still a high probability that many potentially interested parties will not know about the ICO.

The Graphgrail AI project did not provide a financial model, which prevents us estimating the projected costs of maintaining the operating activity, or assessing the degree of dependence of the project’s viability on the amount of funds raised during the ICO.



The information contained in the document is for informational purposes only. The views expressed in this document are solely personal stance of the ICOrating Team, based on data from open access and information that developers provided to the team through Skype, email or other means of communication.

Our goal is to increase the transparency and reliability of the young ICO market and to minimize the risk of fraud.

We appreciate feedback with constructive comments, suggestions and ideas on how to make the analysis more comprehensive and informative.