[DPRating] Cortex: Deep Integration of AI and blockchain

Published in

DPRating

16 min readMay 22, 2018

Cortex (CTXC)

66.4 | B- (investment grade)
Risk: medium
Popularity: medium
Potential: medium (five grades: extremely low, low, medium, high, very high)
Website: http://www.cortexlabs.ai

Advantages of Cortex

Cortex further integrates AI into blockchain by adding the former directly to the latter and expands the capacity of smart contract and blockchain (advantage of original creation).

Disadvantages of Cortex

According to its actual progress, Cortex will roll out test network in the first quarter of 2019. Compared with its rivals, it starts relatively late.

If any of the following occurs, projects may receive a higher score:

Test network for mining of the third quarter is released as scheduled or in advance, or test data meets or exceeds expectations.

In these following occasions, projects may receive a lower score:

Significant bottleneck turns up in the development of AI-based smart contract technology (core competitiveness will be substantially reduced )

Brief:

Project: it positions in AI + dedicated public blockchain, aiming to create a decentralized AI-based autonomous system.

Strategic positioning: 39, Project necessity: 31, Final score: 70

Team: it has plenty of developers with strong academic background of AI and some experience in blockchain development.

Prof. Whitfield Diffie (the recipient of Turing Awrad), Zhou Shuoji (partner of FBG Capital) and Zhou Yahui (president of Kalends) are the advisors fo the project.

Technical team: 34, Operation team: 31, Advisors and investors: 19, Final score: 84

Progress of implementation: according to the official website of project executor, Cortex will be listed on several mainstream exchanges in the first quarter. And it aims to release test network for mining in the third quarter. Test network of AI-based smart contract is planned to be rolled out in 2019. At present, it is still in the early stage.

Under our evaluations, it is now in the small-scale test stage, score: 40

Community governance: The amount raised by Cortex is slightly higher than its rivals. And distribution of tokens is highly matched with descriptions in white paper. The efficiency of token incentive mechanism is not high, token is highly necessary for the project. The possibility of the unlock of token in advance exists.

Funding and budget disclosure: 9, Fund supervision: 20, T0ken appreciation logic: 25, final score: 54

Popularity:

Facebook

183 followers, 2 tweets in April, 0 likes, 0 reposts, 0 comments (low activity).

Twitter

2,459 followers, 29 tweets in April, 146 messages in total (5.03 on average), 745 likes in total (25.69 on average), 310 reposts in total (10.68 on average). According to the data, the quality of activity is ordinary.

47,223 people, 351 people chatting within 2 days (175.5 on average per day, activity: 0.371%).

In summary, Cortex’s popularity at present is medium.

Risk: risk is low in policy and team change, medium in contract fulfilment and competition, but high in capital.

We therefore determine that the risk of Cortex is medium.

Independent dimension: Potential

Evaluations of project’s potential will be made from three aspects — current market value, liquidity and project phase.

Liquidity:

Cortex has been listed on Huobi, OKEx and other 8 regular exchanges. It is ranked №24 by trading volume within 24 hours, which proves its liquidity. However, it does not be listed on Binance and other international exchanges.

Market Value:

When Cortex first came online, the whole market was on the rise. Its price of public offering was twice times higher than average price (0.58 USD). In the current stage, its circulation market value is 269.71 million USD (1.699 billion RMB) which is ranked №75. It makes up less than 0.1% of global circulation market value. Currently average market value of the top 10 projects in competitive field of other classes is about 891.86 million USD (reference: DPC Index). On the basis of its quality, Cortex’s market value is more likely to soar again in the future.

Project Phase:

According to the official website, Cortex will be listed on several mainstream exchanges (e.g. Huobi and OKEX) in the first quarter. Expectations of investors in the future include:

Test network for mining which will be released in the third quarter
Test network of AI-based smart contract which will be released in the first quarter of 2019. There is no method to verify its actual progress and results.

Though Cortex’s AI-based smart contract is very ambitious in terms of technology, its development cycle is long and with great difficulties. Therefore, our evaluations on Cortex in the dimension is medium.

Dimensions:

Project (weight: 30% score: 72)

Brief introduction: Cortex aims to create a decentralized AI autonomous system.

Strategic Positioning

Full score 60, actual score 39

Cortex further integrates AI into blockchain by adding the former directly to the latter and expands thus the capacity of smart contract and blockchain.

Cortex also provides advanced AI learning model on blockchain. It allows users to release tasks on the platform and submit AI DApps (AI Distributed Applications Program) by creating an AI model platform. Excellent AI model will be rewarded when used, which reaches a win-win situation between model developers and users.

1. Competitive field: AI + dedicated public blockchain

Cortex aims to provide one-stop solution to address problems for one field and integrate blockchain technology with AI.

Compared with centralized AI, decentralized AI is of higher degree of freedom. It helps more excellent models to stand out and avoids manipulated results under centralized AI via market competition of models.

2. Innovations:

Cortex brings forward the concept of AI DApps (AI Distributed Applications Program). It introduces AI into smart contract, making the following applications possible:

Information services: system of personalized recommendation, search engine, news writing and auto-induction etc.
Financial services: credit scoring and intelligent investment consulting.
AI assistant: automatic question answering, industry knowledge map, voice modifying and face attribute prediction.
Simulation environment: automatic driving, Go and other applications of reinforcement learning.

Cortex allows users to write programs of machine learning on blockchain and to submit interaction that depends on other contracts. This is Cortex’s original creation. However, whether ‘other contracts’ here refers to smart contracts on ETH or those on different blockchains is not explicit in white paper. If they refer to the latter, it is quite hard to realize, because current technology could not realize the cross-chain transfer of smart contracts. There’s still a long way to go.

3. Competitors

Cortex’s arch-rivals are as follows:

Deep Brain Chain (DBC)(public offering on December 15th, 2017) — an AI-based computing power platform on the basis of blockchain technology.
Bottos (BOT)(public offering on December 31th, 2017) — a decentralized protocol for registration, offering, transfer and transaction via point-to-point network, realizing AI data and model based on blockchain technology.

Although existing rivals start to operate earlier, Cortex further integrates AI into blockchain by adding the former directly to the latter and plans to expand the capacity of smart contract and blockchain. Compared with reinforcement at the computational level, Cortex has bigger imaginary space for potential. However, its difficulty in realization can not be neglected.

Given the above, Cortex does not have first-mover advantage, but with the aid of innovation, it still boasts some certain competitive advantages.

Therefore, Cortex receives these scores in the dimension of strategic positioning:

Project Necessity

Full score: 40, actual score: 31

1. Connection with blockchain: strong

Data layer: blockchain of Cortex is not for the storage of models and data, but for the storage of their hashed values. Key-value storage system is not on blockchains. When there are enough copies disseminating in the network, blockchain of Cortex will obtain new models and data.

Network layer: the implicit cost of Cortex’s operation is Enodorphin which is in charge of analyzing the core instruction set of machine learning and deep learning and is used for communications between Cortex and Block. However, it is not involved in the communications of network layers between Block and Block.

Consensus layer: it uses PoW to further narrow the gap between speed-up ratio of CPU and GPU.

Incentive layer: Cortex comes up with an incentive mechanism of collective cooperation which allows everyone to submit and optimise models. In the meantime, model developers will be also rewarded.

Contract layer: Cortex will add an inference instruction to smart contract, which enables the use of Cortex blockchain in smart contract.

Encryption layer: not mentioned.

Application layer and platform layer: Cortex also aims to create a machine learning platform which allows users to release tasks on the platform, submit models and create their own AI DApps (AI Distributed Applications Program) by inference via smart contract transfer.

In summary, Cortex has a close connection with blockchain in data layer, network layer, incentive layer, contract layer, application layer and platform layer, but its relation with blockchain is not mentioned in encryption layer.

2. Market scale: large

In the current stage, AI is not extensively applied in reality. But it has a broad prospect. Cortex aims to add AI function to smart contract, further expanding thus the capacity of smart contract and blockchain and adding to them the market space of stock and increments.

3. Demands: above the average

With the popularization of smart contract in use, the demand for more abundant and more advanced functions of it will be stronger. Therefore, we determine that the market demand of Cortex is large.

However, the level of demand of Cortex in use cases (i.e. information services, financial services and AI assistant) is medium.

The output of applications of reinforcement learning can be achieved by model of environmental prediction. It is the specific effect of blockchain with the aid of AI technology. Compared with centralized AI system, Cortex’s decentralized system boosts the efficiency and reduces the costs at the same time.

To sum up, we hold that the overall performance in project:

Team (weight 30%, actual score: 84)

Technical Team

Full score: 40, actual score: 35

The CEO of Cortex studied subsequently in Tsinghua University, Carnegie Mellon University and UCSC and became then the chief research scientist of SFTC. Besides, he was the vice president of Coinlark in 2013, in charge of the running of Bitcoin ATM. Moreover, he also established waterhole.io which provided services of mining pool, wallets and sales of mining machine. He is proficient in algorithm and Bitcoin ecology. In the year of 2014, he founded Suishi Tech. In the shareholder change of this company, we found that Beijing Kuailian consulting centre and Cheliu Electronic Tech Limited Partnership in Suzhou Industrial Park were newly added in October and December respectively in 2017. Legal person of the two companies are Zhou Shuoji and Wu Jihan respectively whose backgrounds are in conformity with Cortex’s investors. See figure below:

In the official website of Cortex, another four developers of the team are also introduced. They are Wang Weiyang (CTO), Yang Yang (Chief engineer of blockchain), Xiao Yan (Chief engineer of deep learning) and Tian Wentao (senior engineer of AI and blockchain). All boasts strong academic background. They are respectively master of statistics in University of Chicago, master of computing in Tsinghua University, PhD in computer science of Tsinghua University and master of engineering administration in CUHK. Besides, Wang Weiyang once worked in Wanda Research Institute of Network Technology, in charge of the research into blockchain and machine learning. See figure below:

In addition, Cortex also has a well-experienced community development team who has already achieved a great deal in AI field. See figure below:

In conclusion, Cortex team’s experience in blockchain development is limited in mining, mining pool and bitcoin ATM; however, it has already achieved something in machine learning and boasts many professional talents.

Therefore, our evaluations on Cortex’s technical team are as follows:

Operation Team

Full score: 40, actual score: 31

In official website, we only find information of Amy Chen (chief operating officer) in Cortex’s operation team. In 2012, she co-founded ASOLAR INC (a corporation of solar energy project development) in New York, and founded then UrbanLab Corp (a corporation of consulting and management of solar energy projects). Therefore, she has a certain of operational management experience.

The efficiency of dedicated chips in blockchain field and universal chips is of vast difference. Cortex’s subject of operation — Beijing Suishi Tech gets investment of Bitmain (the biggest supplier of mining machine), which hopefully will be conducive to its development of chips for mining machines.

Here are our evaluations of Cortex’s operation team:

Investors (and advisors)

Full score: 20, actual score: 20

Advisor team of Cortex has elites in all walks of life. Whitfield Diffie (the recipient of Turing Award in 2015 and inventor of public key cryptography) is its academic advisor, Zhuo Shuoji (founder of FBG Capital) assumes the office of strategic advisor, Zhou Yahui (CEO and president of Kalends — an A-share listed company) holds the post of business consultant and Shen Heting (former CEO and president of China Metallurgical Group Corporation) takes the position of policy advisor.

According to official website, investors of Cortex includes several leading investment institutions in blockchain industry and traditional venture capital field, such as FBG Capital, Danhua Capital, Bitmain and ZhenFund.

Here are our evaluations on the investors (and advisors) of Cortex:

Actual Progress (Weight: 10%, actual score: 40)

Roadmap released at Cortex’s official website is as follows:

According to the official website of project executor, Cortex will be listed on several mainstream exchanges in the first quarter. And it aims to release test network for mining in the third quarter. Test network of AI-based smart contract is planned to be rolled out in 2019. At present, it is still in the early stage.

Cortex’s library on github is https://github.com/CortexFoundation. Currently, but there are not many updates in codebase. It mainly updates an example to demonstrate Cortex’s structure and Cortex Competition of Cortex labs. See figure below:

Given the above, we determine that the project is still in the DEMO phase with a score of 40.

Community Governance (Weight: 30%, actual score: 71)

As for projects in embryo, we focus on its system of community governance including financing volume, capital supervision and whether they have self-consistent closed loop logic to promote the appreciation of tokens.

Funding and budget disclosure

Full score: 30, actual score: 9

Details of this year’s public offering from February 7th to March 7th are as follows:

Conversion ratio: about 1ETH = 1500CTXC

Raised amount: 60,000,000 CTXC (about 20.01%)

Funding: Approximately 34.5948 million USD, if calculated by ETH’s average price on closing quotation from 7th February to 7th March which was about 864.87 USD (reference: CMC).

Distribution of tokens: miners (rewards for mining): 50.03%, investors (Genesis Block): 20.01%, project foundation (Genesis Block): 24.95% and advisors/scholars/community (Genesis Block): 5.00%.

Overall market cap: 172.974 million USD.

In January 2018 Matrix (MAN) a project aimed to improve smart contract users’ experience with the aid of AI. It raised approximately 12.59 million USD and sold 21.73% of tokens with a valuation of 57.98 million USD.

DeepBrain Chain (DBC)(an AI-based computing power platform on the basis of blockchain) raised 74.34 million RMB (approximately 11.80 million USD) and sold 15% of tokens in the public offering on 15th December 2017. Its valuation was 78.368 million USD.

In terms of financing volume, Cortex though has a bigger pattern, its financing volume is just slightly bigger than similar projects.

Budget disclosure is not mentioned in white paper.

Therefore, we conclude in this dimension that Cortex’s financing volume is slightly bigger than similar projects, but its budget disclosure is vague. Full score: 30, actual score: 9.

Fund Supervision

Full score: 15, actual score: 7.5

After fundraising, whether projects can have an efficient and transparent system of capital supervision is also an important indicator to judge their development in the future. However, capital supervision is not mentioned in Cortex’s white paper. Hence, Cortex’s score in this dimension is 7.5.

Distribution of Tokens

Full score: 15, actual score: 13

Tokens distribution is highly matched with the plan released officially. Therefore, with a full score of 15, Cortex gets 13.

Token Appreciation Logic

Full score: 40, actual score: 25

In the crowdfunding of blockchain projects, what retail investors get is equity units (i.e. tokens) released by project executor. Therefore, it is one of the core factors of projects’ investment evaluations that whether a closed logic loop for token’s appreciation and incentive can be formed in project’s design of community ecology.

According to the economic design of tokens in white paper, Cortex mainly considers encrypted blocks in mining and reward&competition mechanism of model developers. Tokens are mainly used for the payment of block and model used. On top of them, Cortex also adds the concept of Endorphins in order to calculate the charge of computation of Cortex’s virtual machine on the level of graphics card.

As for projects of public blockchains, their core is DApp’s ecology in the future. Therefore, we change our standards of evaluations on token’s appreciation logic. We’ll evaluate from two aspects — projects’ connection with tokens and efficiency of the token appreciation logic.

Projects’ connection with tokens refers to the connection between the value of digital tokens released by the project executor and the development of project.

Efficiency of token’s incentive mechanism refers to the clarity, effectiveness and directness of token’s appreciation logic in the system described in white paper.

On top of those, the competitiveness of projects, their technologies and team’ background resources will also affect the prospect of public blockchain projects’ ecological construction.

1. Efficiency of token’s incentive mechanism: CTXC Token’s design of internal incentive mechanism is quite simple. Other than functions above-mentioned in official white paper, there is no effective appreciation mechanism for tokens, such as buy-back and token destruction. Therefore, we determine that Cortex’s token incentive mechanism is commonly efficient.

2. Projects’ connection with tokens: Cortex mainly considers encrypted blocks in mining and reward&competition mechanism of model developers. Tokens are mainly used for the payment of block and model used. Therefore, we determine that Cortex’s connection with tokens is strong.

3. Additional offering and lock-up plans: in white paper, Cortex only mentions the locked position for Project Marking Lock. Details of other tokens’ locked position are not disclosed, but the possibility of a rapid unlock of the tokens exists.

Aggregate amount of tokens is fixed at 60,000,000 CTXC

Given the above, we determine in this dimension that Cortex’s incentive mechanism for tokens is commonly efficient and this project is of strong connection with tokens. Rapid unlock of most tokens is not mentioned, but this possibility exists. Full score: 40, actual score: 25.

Here are our evaluations on Cortex in this dimension:

Independent Dimension: Risk

As for projects in embryo, we focus on these several aspects while evaluating risks:

According to our evaluations, Cortex is faced with the following risks:

1. Policy risk: low

Government’s regulation in respect of legal compliance and regulatory can rarely affect AI.

2. Team change risk: low

It has plenty of team members with high transparency that all members’ information at official website can be checked at LinkIn. Their has been no major changes to the team. .

3. Contract fulfilment risk: medium

The time span of project’s roadmap is quite long and there’re great difficulties in the development of technology. Therefore, we determine that it has medium potential risk in contract fulfilment.

4. Competition risk: medium

Although existing rivals take the lead in operation, Cortex further integrates AI into blockchain by adding the former directly to the latter and plans to expand the capacity of smart contract and blockchain. Compared with reinforcement at the computational level, Cortex has bigger imaginary space. However, its difficulty in realization can not be neglected.

5. Capital risk: high

Information of team position and locked position is not mentioned in any disclosed open information.

We therefore determine that Cortex’s risk in this dimension is medium.

Independent Dimension: Popularity

When evaluating Cortex’s popularity, we mainly looked at its official social media, such as the activity of its telegram group and slack group. At the same time, we also take into account the number of followers it has on Twitter and Facebook.

Up to 18:00 on May 7th

一、Facebook：

1. Facebook:

183 followers, 2 tweets in April, 0 likes, 0 reposts, 0 comments (low activity).

二、Twitter

2. Twitter

三、Telegram

3. Telegram

47,223 people, 351 people chatting within 2 days (175.5 on average per day, activity: 0.371%).

In summary, Cortex’s popularity at present is medium.

Here are our final evaluations after summarizing the results from all dimensions: