As cryptocurrencies enter their second decade of existence, the ideas and problems developers are trying to address become more and more complicated. The potential this new industry has to change the way our day to day lives function is clear for everyone to see, it just comes down to fulfilling said potential by learning how to apply the mentioned ideas practically.
One idea that recently saw the light of day could potentially solve problems in the areas of AI, IoT, transportation, and worldwide economic activity, among other things. Sounds impressive and ambitious, doesn’t it? In today’s article we’ll be analyzing the project behind this idea, called Fetch.ai, and determining if there’s any substance behind what they are trying to sell us.
Fetch.ai boasts an interesting batch of seasoned businessmen, developers, college-educated individuals and crypto investors/enthusiasts.
Educational institutions such as University of Oxford, University of Amsterdam, University of Birmingham, University of Miami, UCL, King’s College London, as well as former places of employment such as Swiss National Bank, BBC, Sony, Citrix, Hitachi, Google, should guarantee the experience and qualifications of Fetch.ai employees. Some of them include:
Humayun Sheikh (CEO, Co-founder) – An innovation entrepreneur, founding investor in DeepMind. DeepMind was a major AI development company which managed to create the famous AI which defeated the world’s best Go player. The company also developed some impressive chess/other AI’s and was eventually sold to Google for $700 million. Humayun self-reportedly has a record in revolutionizing trading in steel sector and now intends on changing the way we transact, travel, spend, and more.
Toby Simpson (CTO, Co-founder) – Producer of the successful a-life Creatures series of games and early developer at Deepmind. His thirty years of experience in software (of which ten were spent as a CTO) are now focused on crypto-economics and Fetch.ai.
Thomas Hain (CSO, Co-founder) – Professor at Sheffield and established scientist in advanced machine learning AI. His job at the project is to bridge real world and academia and is inspired by the opportunities AI brings to modern society.
Jonathan Ward (Head of Research) – A researcher in machine learning, complex systems and blockchain technology. Excited by the challenge of deploying decentralized multi-agent systems in smart cities, supply chain and healthcare. PhD in Machine Learning from UCL.
Troels Ronnow (Software Engineer) – Troels is a young specialist in Blockchain and quantum computation. As a Research Leader in Nokia Technologies, Troels focused on Blockchain application and developed a Bitcoin wallet for Nokia. He is also the co-author of 35 patent applications and apparently has deep expertise in AI algorithms.
Khan Baykaner (Lead Software Engineer) – Research engineer in machine learning with experience developing deep reinforcement and generative models, solving problems in digital health, audio/video media and NLP. PhD developing computational auditory models from University of Surrey.
Still, these past credentials and recommendations don’t mean that the project will definitely succeed. It’s somewhat apparent that most of the people involved in this project don’t have much blockchain development experience, at least not one that can be publicly confirmed and measured.
Humayun is the biggest question mark here, as we couldn’t find public records of his connection to DeepMind. Additionally, Humayun’s other companies seem to be operating with massive debts which might indicate that the current CEO of Fetch.ai isn’t that good at managing funds (or operating companies, for that matter).
Finally, most of the project’s developers don’t have a large public development profile/print, meaning that it’s extremely hard to determine how good they are (and if they are real developers at all).
Fetch.ai is currently a member of the MOBI, a collaborative effort working on developing blockchain-powered vehicle data/mobility services. MOBI (Mobility Open Blockchain Initiative) is seen as a very ambitious and prestigious movement that also has names such as BMW, Ford, Bosch, General Motors, Renault, Accenture, IBM, and some crypto projects such as BigchainDB, Hyperledger, and IOTA on board. MOBI will be cooperating with Fetch.AI on projects related to:
- Vehicle identity, history and data tracking
- Supply chain tracking, transparency, and efficiency
- Autonomous machine and vehicle payments
- Secure mobile commerce
- Data markets for autonomous and human driving
- Car sharing and ride hailing
- Usage-based mobility pricing and payments for vehicles, insurance, energy, congestion, pollution, and infrastructure.
Outside of MOBI, Fetch.ai has joined forces with enterprises such as Outlier Ventures, TokenMarket, AiiN, Blockchain for Europe and Uledger.
There’s quite a lot of hype around Fetch.ai, that’s for sure. Publishing a short and somewhat braggadocios introductory Medium post back in 2018 which simply said “We’re the world’s first smart ledger” was more than enough to tickle the imaginations of many crypto enthusiasts. Their imaginations were further enticed by a pretty strong marketing effort that was made to promote the project ever since it originally launched.
Just type in “Fetch.ai” in YouTube and you’ll find several well-produced videos that market the project and attempt to hook in potential investors. Additionally, the project was featured on pages of mainstream publications such as The Telegraph, The Economist, The Guardian, Business Weekly and TechCrunch, which certainly helped raise its public profile.
And while the hype is certainly strong, this doesn’t necessarily translate that well into the real world performance or investment. The project is still far from having a fully working product. They promised to have a testnet ready before the ICO; said testnet was launched with a slight delay (as was the ICO). That doesn’t have to mean much, as plenty of legitimate projects slightly overshot their deadlines and had rocky starts to their crypto journeys. Fetch.ai did explain their delays, citing the worsening market conditions and desire to prepare the infrastructure properly as their main reason for missing their set timelines.
The technology side of the project is where we get into the meat of the matter. Fetch.ai wants to create a hybrid of blockchain and AI technologies, one that would have the world’s first self-adaptive and self-regulating decentralized ledger manage people’s transactions. Their model is structurally divided into three elements:
Autonomous Economic Agents — AEAs
These are the “digital citizens” of the Fetch.ai network which will be in charge of receiving and using the network’s digital data. Basically they will act as fully-abled AI representatives of individuals, businesses, organizations and even IOT devices; AEAs will collect the data and then sell it to those agents who want it.
The entire interaction is automated and done through 5 key steps: search and discovery, communication, negotiation, collaboration, and execution/trust. As artificial intelligence usually does, AEAs will learn as they go, meaning that each interaction/mistake they make will help them become better at what they do.
The project expects to see its AEAs transform all the industries that require data, transportation and energy; there market potential is huge as there isn’t a single industry that doesn’t heavily rely on at least one of these.
Open Economic Base
Each agent’s behavior will be “governed” by the Open Economic Framework (OEF), a combination of APIs, directories of services and agents, previous transactions, wallets and agent positions. It is designed to present agents with semantic, geographic and economic views on the world.
OEF will store relevant information and use artificial intelligence to optimize the use of that information for predictions/support for the AEAs. OEF will be self-adaptive and will depend on special “trusted” nodes that will receive token rewards for providing their services.
Fetch.ai’s Smart Ledger will be a combination of the traditional blockchain architecture and the DAG (direct acrylic graph) technology that was presented by the IOTA project.
Many are comparing it to Zilliqa on the basis that it will have sharding; unlike traditional sharding, in Fetch’s Smart Ledger we have so-called resource lanes. A transaction may be assigned to several different lanes simultaneously.
The system will shard in a manner similar to chain forking, where new lanes can be created to ease off the pressure on a cluttered transaction pool. Old lanes will be referred to by two new ones, creating a DAG-like structure which allows for blockchain data consistency.
Fetch.ai can form a transaction block by referring to the previous block hashes in the groups that are defined in the transactions resources. Overall it’s a complicated system aiming to deliver unparalleled scalability, stability, useful economic work and quality information which you can learn more about by reading their technical whitepaper or their Fetch.ai Ledger Yellowpaper.
Fetch.ai will utilize what they call Useful Proof of Work (uPoW) consensus mechanism which combines the elements of PoW, PoS and DAG.
The DAG system will deem any transaction valid once it is confirmed by two nodes, leaving computational resources free to train the AI. Additionally, low power nodes can earn blockchain rewards by validating low value transactions. They also talk a lot about machine learning and AI, which will be heavily featured in the entire protocol. Methods such as process mining, long short-term memory, and recurrent neural networks will be built into the protocol to ensure authenticity/ optimal performance and allow for involved parties to trust the network.
Fetch.ai token uses the ticker FET and is your typical ERC-20 launched through a token generation event. It will have a total issued supply of 1,152,997,575 tokens (further divisible), with the distribution being handled as follows:
- Foundation – 19.99%
- Founders and Team – 19.99%
- Public ICO – 6%
- Private sale – 6.38%
- Seed sale – 5.24%
- Advisors – 10%
- Mining Rewards – 15%
- Future Releases – 17.4%
You can check out which ones will have any lockups or other special conditions attached to them here. During seed/SAFT private sales, just a bit more than 11% of tokens were sold to early investors. Those funds were spent in the following way, according to the team:
- Team – 25%
- Development resources – 45%
- Partnerships – 10%
- Marketing – 10%
- Professional Services (whatever that meant) – 5%
- Miscellaneous – 5%
The tokens of the advisors and founders will be unfrozen gradually, through a period of three years. Public sale of FET tokens begun on 25th of February and lasted only 10 seconds, as this was enough time to sell the entire private sale supply of 69,204,152 FET. Investors were required to use BNB tokens to purchase FET. The token was sold at a price of $0.0867 per token, with a minimum purchase of $20. The Fetch team also set a maximum personal purchase cap of $3,000 to ensure more people are able to get a piece of the pie.
The project raised additional $6 million through this ICO (which was Binance Launchpad’s second project), confirming that interest for public coin offerings certainly hasn’t waned as much as many crypto naysayers would like you to believe.
The token will play a crucial role in the Fetch.ai network, driving its economy and data exchange. Agents will need FET to pay for data, access services, and infrastructure, develop algorithms etc. It will also be used as “gas” on the Fetch.ai network.
Fetch.ai As Invesment
Fetch.ai remains somewhat secretive about how exactly their technology will work, as if to prevent competitors from borrowing their ideas.
This is why they decided to not have their smart contract open source but have instead hired a company called Hosho to privately audit the code. The Fetch.AI smart contract passed a wide range of tests using the Meadow testing framework which some might find surprising, as some minor code tweakings are usually expected after audits.
Hosho has reviewed over 100 smart contracts and has a reputation of being knowledgeable on the topic of Solidity smart contract vulnerabilities, so they are to be trusted. Still, having a public audit is usually the best litmus test for coders.
This isn’t the only place where they chose to remain secretive in the past. Most of the coding efforts were done through private GitHub repositories, in C++ and Python.
The project does have a public GitHub as well which was updated periodically but coders focused mostly on the private ones. Overall, the project’s front page claims that over 100 thousand lines of code were written for the Fetch.ai so far.
According to Binance’s Launchpad analysis of the project, from February 2019 onwards, Fetch.AI will move all primary development activity to the public repositories “to better interact with the wider development community and to ensure that a more regular set of updates and commits are available to all.”
The development is pretty much ongoing and the project is currently in the “private test network” stage of its roadmap. Further down the line we can expect to see first improvements to the consensus and OEF (Q2), with the team hoping to include multi-dependency auctions and enhanced decentralized ledger computing into the mix.
By Q3 of 2019 the team expects to release its alpha/beta networks, with final mainnet launch expected in Q4. Keep in mind that what they are trying to do with their technology is rather experimental and these timelines could easily end up being broken due to unforeseen reasons.
Fetch.ai is a project that managed to garner a lot of hype prior to its ICO, mainly because of their quality marketing efforts and the fact that they had the support/backing of Binance’s Launchpad. One wonders if the project was overhyped and if this perhaps led to its token having too much speculative value.
We saw something similar with BTT token, whose price managed to shoot up to 12x from the private sale values merely on hype/Justin Sun’s army of bots shilling the project into oblivion. BTT’s price hasn’t been faring that well since those highs, suggesting that the naysayers who questioned both the hype and the token’s reason for existing might have been correct.
To not stray off the topic too much, Fetch.ai does draw some parallels to Binance Launchpad’s original project but is also quite different in many areas.
FET token did jump 400% post its ICO and has since lost more than half of its ATH value; this drop suggests that there was plenty of speculators involved who sold off their tokens immediately once they received them. This price drop could continue, as there is a significant amount of tokens currently in lock-up or is yet to be released to the public through mining and “release events”.
This is where the similarities between the two projects end. Fetch.ai comes with a much more feasible use case, even though it involves AI and meshing of complicated blockchain technologies (this perhaps tells us more about how useless BTT token actually is than about the feasibility of Fetch.ai’s idea).
If the people behind the project stay true to their mission, Fetch.ai could signal the era of commercial blockchain application and adoption. Still, you should be careful; what they are trying to do is equivalent to sticking a square peg through a round hole. There’s much work to be done here and his one definitely seems like a long-term project.
Be aware that they might easily come across some difficult road bumps along the way, including lack of expertise/financing that could ultimately result in the project’s vision not coming true. Combining this fact with the abovementioned price inflation issues leads us to believe that it might be best to sit investing into Fetch.ai out, at least for the time being.