Agrello CEO: AI Can Help Us Understand How to Make Better Contracts
Legal innovation is one of the most noticeable trends now. Thanks to the advancements in blockchain technology and machine learning, it is now possible to effectively fuse the benefits of both and build systems to create and execute contracts that are literally smart.
At the Consensus 2018 conference in New York, lawless.tech’s Tamar Menteshashvili talked with Hando Rand, CEO and Co-founder of Agrello, an AI-driven smart contracts platform, about the applicability of AI in a smart contract ecosystem and the latest innovations in smart contracting. Mr. Rand has a background of work with international law, escrow and automation. Currently he is busy with the Agrello project seeking to build an ecosystem of smart contracts and intelligent software agents for everyday needs with features such as virtual passport, legally binding digital signature, template library, and community developed apps for contract execution.
lawless.tech: How big is your team? How and why did you come up with the idea of self-aware contracts? Were you inspired by Ethereum smart contracts?
Hando Rand: Our team is 40 people. I was somewhat inspired by Ethereum. Once it became public I saw their [smart contract] concept and, as I have a legal background, thought that it would have problems working as a legal agreement. I had my own ideas on how to make a machine-readable contract, but my approach was completely different. As I was a big IT guru with legal background, I understood the structure of a legal contract, the most important element, very well. I’ve noticed that lawyers write differently: they use different sentence structure, different general formatting of a contract. However they all do write similar things and include the same elements. A standard way to put these legal constructs together into a contract would easily make such a contract machine-readable.
lawless.tech: Is combining AI and smart contracts an essential phase of technological advancement or just a nod to recent trends?
Hando Rand: It is super essential, as AI can help us understand how to make better contracts and what to expect from people when they perform contracts. When contracts are being executed en masse, AI as well as people can learn how to provide many essential services not only to the contract system itself, but to the economy as a whole.
lawless.tech: What are the functions of AI in your system? How did you train it?
Hando Rand: Our AI is not ready yet. We expect it to provide business intelligence, general market analytics, current trends. Based on the big data about the previously detected patterns in contract parties’ behaviour, it will also provide analytics predicting how a party would act in certain circumstances. This would give us additional tools to create better contracts and assess the risks entailed by the obligations stated in such contracts.
lawless.tech: How did you implement the “text-to-code” compilers? How do you think the mass adoption of machine-readable contracts will happen? How long it will take?
Hando Rand: A contract is a construct of obligations and rights. This applies to any jurisdiction. A jurisdiction defines what can be written inside the contract as an obligation, but it doesn’t modify the fact that there is an obligation. We standardise an obligation construct (this is also explained in our Whitepaper, section 3). Obligation and rights can be taken into smaller pieces, as basically every word or two play a certain role in the contract. The information is already structured in human language so it can be automatically compiled into a machine-readable declaration and even humans will be able to see how it works.
Mass adoption will take a while. Currently, even though contracts can be digital, they are just dummy text files. A lot of time will pass until bigger companies adopt this new technique, but it will definitely happen.
lawless.tech: Will the transition from classical written contracts to smart and self-aware contracts become a trend?
Hando Rand: We are starting our first products this year. To become a trend it will need to show a few very well implemented and impactful use cases. We try to work with big corporations. I believe that mass adoption won’t happen until 2020. Before it is still an adoption and development period. But I also believe that this year we can already create some traction.
lawless.tech: Currently smart contracts don’t have the same legal force as written contracts in any jurisdiction. How do you solve this problem and what solutions are on the table in general?
Hando Rand: First, our contracts are written contracts, but they are written in such a way that they can be compiled into a machine-readable code automatically. Second, we already have an electronic identity and an electronic signature to verify the parties involved in a contract. Ethereum does not have this feature. The main elements of a contract in any jurisdiction are offer and acceptance. Both should have a declaration of intent, which is usually a signature, but it can also be an act that shows that there is an agreement I want to be bounded with. An offer should have a substance. Otherwise it is not clear what do the parties consent to. In case of Ethereum smart contracts, it is very hard to understand what is offer being offered or what responsibilities are exchanged.
lawless.tech: Written contracts usually contain the parties’ personal details such as addresses or credit card numbers. How is the sensitive and personal data protected during the transition from a written contract to a smart or self-aware contract?
Hando Rand: Basically, the database structure does not differ too much from what it is today. We usually use local databases or cloud storage. There are slight differences. The main thing is that we don’t store any sensitive data in the blockchain. We store only the hashes which mean nothing to people who are not involved in this relationship. The sensitive data are stored in the cloud or users’ local databases. We are pretty much database agnostic, as our tech can be implemented with different databases. Our public platform will work using the cloud. We are currently working on homomorphic encryption techniques, meaning, basically, that everything in the cloud is encrypted and can be decrypted only with Agrello ID. Only the signing parties will be allowed see this. But agents and AI can execute the functions with the information of the contract while it is encrypted. This technique is similar to what is generally referred to as “zero-knowledge proof.”
lawless.tech: Is there a possibility of data misrepresentation while concluding a self-aware contract? For instance, if a party submits fake documents while concluding a written contract that is afterwards turned into a self-aware contract, the SAC will contain false information while being valid. Is it an existing problem or just a theory? How do you tackle this kind of a problem? Will it be possible to terminate the contract in case of violations?
Hando Rand: We use verification techniques with high fraud detection level, so signing with a fake ID is unlikely to happen. If there is a fraudulent representation of an actual identity, the dishonest party is still involved in a contract despite the incorrect identity. It would be difficult, but possible to track back such dishonest person. The fraudster would remain a party in the contract anyway, even if they would use fake ID.
lawless.tech: The funds transferred between wallets can’t be returned without the involvement of the receiving party. Is it possible to return the funds that were transferred to the other party in accordance with the SAC terms?
Hando Rand: In this case we can use a deposit mechanism combined with our smart dispute resolution system. The arbitrator who resolves a dispute decides whether to release the deposit or hold it. Depositing is actually very simple with Agrello. You just have to add an obligation telling that a certain party has to pay the deposit and then an obligation to return it later on. This would create a deposit function in Agrello SIC. However, if there is no deposit requirement and something goes wrong, one party will have the right to make the contract void, but it would more difficult to return the sent money, as the recipient may disappear. We can link the wallet to the obligations inside the contract, though, unless there is a deposit, it is not possible to restrict the funds from being taken away.
lawless.tech: Why do you need a specially designed programming language for your network? How was the Agrello Language created?
Hando Rand: Being somewhat aware about programming, I looked into what is a contract and what should be achieved with smart contracts. It would be very nice to have automatic execution for smart contracts. However, it is impossible and might bring a lot of problems if there is only an automatic execution procedure. So there must be some kind of a hybrid.
I digged into the core of a contract, which is obligations and rights, and noticed that the Turing-complete smart contract approach is not what we want. Generally, anything that is designed to fit everything is actually good for nothing. So we designed our smart contract language to be a declarative language set so that the code of the contract is just machine-readable declarations and there are no actual functions inside the contract. This makes the composition of those contracts way easier than with Solidity in case of Ethereum. We can pre-program all the agents dealing with this information and the AI functions. We don’t really have to change them or we can update them if needed,while not changing the contract at all. Also, we can keep the functioning AI parts working and change the contract all the time. With Ethereum’s Solidity you would have to make the functional and declarative parts again and again, building it every time from scratch, which makes dealing with smart contracts very difficult.
lawless.tech: In your blog post you’ve mentioned that eventually Agrello may or will be ported to other blockchain platforms, such as Lisk, NEM, or Qtum. How are you going to implement it, considering that said platforms use different programming languages for their smart contracts?
Hando Rand: We don’t depend on smart contract language inside blockchain, as we only use the immutability function of the blockchain. We put hashes into transactions to keep immutability in a smart contracting system. We have the cloud for the rest. We do not depend on the differences between smart contract capable blockchain and non-smart contract blockchain. We already ported Agrello to NEM. Our electronic signatures are working on NEM. Electronic identity currently uses some smart contract features from Ethereum, but, as we go further, we might port it to some other blockchains as well.
lawless.tech: According to the roadmap, in 2018 you will attempt to enter the Asian market and achieve several other goals, such as software release, partnerships with cloud storage projects, or aforementioned Agrello port. What expectations do you have of your 2018 considering your 2017?
Hando Rand: We have a very interesting outlook towards Japan. We hope to reach a very important deal there. More information will be available soon. Now we are launching out electronic identity and electronic signature systems and we already have some integrations on that. We are looking to make a pilot use case for our smart contract system. I hope to have the pilot going in Summer. We want to make a very good scalable use case. As for cloud storage, we saw that the problem with decentralised storages is that once you upload the data, you cannot modify it anymore. Users have to send new entries every time. We are looking into the using a properly decentralised storage that can scale.
My most important expectation for 2018 is to see that the electronic signature and electronic ID really work and I will be able to sign contracts with Agrello platform already. The second expectation is that we will be able to show the self-aware contract tech. We also want to open up the legal contract template marketplace. It can start from simple contract templates and later lawyers from all around the world can contribute. It is important to get token mechanics working as soon as possible.
It looks like a matter of time when old fashioned written contracts will go electronic. For now, however, the search for more advanced forms of contracts continues, with the end game being about keeping all the benefits of conventional written contracts while fitting into our inevitably digital future.
The way to the actual legally binding smart contracts is long, but the technology prototypes are already on the table. It seems that we’re standing at the brink of a long adoption period that, as it usually happens with innovation, will depend greatly on the initiatives from tech giants and their willingness to bring the tech to our daily lives.
Follow us on Twitter to stay tuned on the recent developments in regulation of new technologies, and be the first to read expert opinions and interviews.