To put it briefly, RepuX is seeking to build a platform that will allow organisations to sell anonymised data to developers. Those developers will then use that data to train machine learning algorithms to build applications that aid decision making. The platform would also act as a marketplace for those applications.
This is an interesting project in what is shaping up to be a very lucrative market. There are also very few projects operating in this space, and RepuX has put together a compelling white-paper.
CONCEPT AND WHITE-PAPER
Before discussing the white paper itself, it’s worth looking at artificial intelligence and data. Companies and individuals create data whenever a transaction is made and whenever an application or device is used. Most of that data goes unused, and until recently had little value.
The rise of artificial intelligence and in particular machine learning has created a need, and therefore a market, for that data. It is estimated that artificial intelligence development more than doubles every two years. In 2016, companies invested between $26 and $39 billion developing artificial intelligence.
Machine learning is a field of artificial intelligence. In very simplistic terms, machine learning algorithms build knowledge by using historical data. These algorithms are therefore of little use without data.
This is where RepuX comes in. The platform is being positioned to act as the bridge between data collectors and data users. In particular, RepuX sees an opportunity within the SME market. While large corporations have teams of developers and can afford to source expensive data, SMEs have neither the budget nor the expertise to make use of that data.
In addition, machine learning algorithms need datasets that are large enough to be statistically significant. This isn’t a problem for companies like Amazon, Google and Microsoft that generate millions of data records every day. But for most SMEs, sourcing large datasets is an obstacle. These SMEs need to either source data from companies that do collect large amounts of data or collate data from multiple sources.
RepuX’s vision is for an ecosystem which allows developers to buy data from organizations, use that data to build decision-making applications, and then sell those applications to organizations.
The Whitepaper, which is 50 pages long, is professional and comprehensive. It covers the background to the industry, lists examples and potential uses cases, and goes into a fair amount of detail concerning the technology being used.
Some of the more interesting use cases identified include:
- Making it possible for individuals to trade their anonymized browsing data for access to subscription software services.
- Allowing museums and archives to monetize their databases.
- Reducing the cost of data used by natural language processing algorithms.
- Aggregating anonymized medical records from multiple hospitals to build valuable databases.
The platform itself will have several layers including an API, a logic layer managing reputations and trust, and a decentralised data layer. Data will be stored using InterPlanetary File System (or IPFS) protocol.
The team have a pragmatic attitude to the technology they use. The network is currently being built on the Ethereum blockchain, but the white paper emphasises that they are open to using other protocols and even developing their own where necessary.
The white-paper is very upfront about the risks and challenges the project will face. While the potential opportunity is significant, RepuX is operating in a very new and dynamic industry, and it is difficult to foresee all the hurdles they may face in the future. The fact that the team are aware of and transparent about these risks is a positive sign.
RepuX was originally seeded in July 2017. In December the first live version of their marketplace was released. At the time of writing this review, the marketplace was limited and included just four data files. The project is therefore somewhere between the proof of concept and minimum viable product stage.
The token sale will conclude on 5th April, and the RepuX protocol is scheduled to be open for developers in May. In June the RepuX platform is scheduled to be available for Decentralised Enterprise Applications.
The project has come online very quickly and has a tight timeline going forward. Ideally, we would have liked to see more detailed plans for the future.
The white-paper lists 11 team members and 5 advisors. The project is being led by Marcin Welner, who is also COO at WorkHQ, a partner company. In fact, most of the development team also work at WorkHQ.
Several advisors are involved in other blockchain projects, among them:
- Douglas Pike, the co-creator of VeriCoin and an advisor to ChainTrade
- Mateusz Mach, Co-Founder COO at Opus foundation and CEO of Five App
- Dr Jay Best of Cogni Capital
All the profiles connect to active LinkedIn accounts, and some of the developer profiles link to active Github accounts too.
At the time of writing, there are some minor discrepancies between the advisors listed on the website and those listed on the whitepaper. However, those listed above are on both the website and whitepaper.
All transactions on the platform will require RepuX tokens. So, developers will buy data using RepuX tokens and then sell their applications to organizations in return for tokens. Developers will also be required to hold tokens to build and run third-party applications on the network.
Half of the 500 million tokens being created are being sold during the presale and token sale. There’s a 50 percent discount during the presale, which also has a $30,000 minimum.
During the main sale, a sliding discount from 30 percent down to 10 percent will apply. It may be of interest for some to note that unsold tokens will not be burnt, but will be managed by the RepuX Foundation, and sold at a later stage. Further details on this point can be gained by consulting with the project team directly.
Of the remaining tokens, 23 percent will be sold on the platform. That means the project may be able to raise revenue in the future. The founders receive 12 percent of the tokens. It’s nice to see that a three year vesting period applies to the founder’s tokens.
The remaining 15 percent goes to the rewards pool, ambassadors and the bounty program.
Long-term success will depend on successful marketing to developers, those who use data and entities that collect data. Building partnerships will also be critical driving to adoption. The budget allocates 25 percent to marketing and 20 percent to building partnerships.
To date, RepuX has signed up three companies; WorkHQ, FungyuCPA and Dascom, on a pilot program. These three companies have a collective reach of millions of companies, predominantly in China.
We spoke to a member of the team who was well aware of the fact that marketing to SMEs will be essential and will also be expensive. The team foresee it taking three to four years to reach profitability, a period for which they should be adequately capitalized.
The team is currently hoping to partner with a large company. If they can do that, the path to breakeven will be considerably shorter.
At the time of writing the review, the project had a very active Telegram group with just over 1,500 members. The team members participating in the group were happy to answer questions and find answers if they didn’t have them at hand. All indications point to the pre-sale attracting a lot of interest.
The Economist magazine stated that “data is the new oil.” That’s probably true, and this project is positioning itself as a vital cog in the new data economy.
One might question the need for a decentralised model. As far as the data itself goes, a centralised model may work just as well. However, the RepuX decentralised model does provide an incentive to developers to build applications and create an ecosystem around the protocol. That could add a lot of value to the platform.
The project is not without its risks, as the whitepaper points out. It’s difficult to gauge how long it will take for SMEs to get used to the idea of selling data, and its difficult to gauge how big that market will become. To reach critical mass, RepuX will need to create a substantial market.
However, adopting a longer term vision, there is little doubt that machine learning will become an increasingly important part of the economy over the next decade.
It’s also clear that smaller companies will have to leverage the power of big data if they are to compete with large companies like Google and Amazon. RepuX has what looks like a solid plan to position themselves as a vital platform within this space.