What if there were a crypto protocol that was specific in mediation on on-chain disputes?
Imagine that, whenever the prophecy markets such as polymercate settled in a controversial manner, the users had a formal way to appeal through a type of neutral on-chain court system. Or if a skilled, knowledgeable third party can rely on a skilled, knowledgeable third party to help decentralized autonomous organizations (DAO) decide. Or if the insurance contract can automatically execute the payment when specific real -world events occur.
It is essentially manufactured with Albert Castilna Laluis and her team Genlair, a crypto project that brings itself to the market as a decision -making system, or a trust infrastructure.
“We are using a blockchain, which has many AIS coordination and compromise on subjective decisions, such as he was a judge,” Castelana, Yazrai’s co-founder and CEO told the coindsk in an interview. “We are originally creating a global synthetic jurisdiction, with an embedded court system that does not sleep, it is super cheaper, and it is super fast.”
The demand for such a mediation project may increase with the development of AI agents in the coming years – sophisticated programs run by artificial intelligence that are capable of performing complex functions in an autonomous manner.
When it comes to crypto markets, AI agents can be used in all kinds of ways: to trade memecoin, mediating bitcoins on exchanges, monitoring the security of DEFI protocols, or providing the insight of the market through analysis, to give only a few use-mammals. AI agents will also be able to appoint other AI agents to complete more complex assignments.
Such agents can proceed at an unexpected rate, Castelana said. In his view, most crypto market participants can manage one handful of them by the end of 2025.
“These agents, they work super fast, they do not sleep, they do not go to jail. You don’t know where they are. Are they going to pass the money-laundering rules? Do they have a bank account? Can they also use a visa card?” Castellana said. “How can we enable fast transactions between them? And how can there be faith in such a world?”
Thanks to your unique architecture, the genlayer can provide a solution by allowing institutions – human or AI – to get a reliable, neutral opinion to weigh on any decision in record time. “Anywhere where you will normally have a third side made of a group of humans … we change them with a global network that provides a consensus between different AIS, a network that can decide in a way that is correct and as fair as possible,” Castelana said.
Synthetic court system
Genlair does not try to compete with other blockchain such as bitcoin, atherium or solana – or even DEFI protocols such as uniswap or compound. Instead, this idea is capable of connecting gentarer for any existing crypto protocol and using its infrastructure.
The series of genlayer zksync, powered by an Etreum layer 2 solution. Its network counts 1,000 verifications, each connected to a large language model (LLM) such as openiI chat, Google’s Burt or Meta Lama.
Suppose a market on polymarkat dwells in a controversial manner. If the polymercate is connected to the genleir, the prediction market users have their synthetic court systems to create “transactions”, Castailna’s ability to increase it (or, as Castailana said).
As soon as the transaction arrives, Genlair selects five verifications randomly to rule it. These five verifications do an LLM querry of their choice to find information on the subject in hand, and then vote on a solution. Which creates a decision.
But polymercate users, in our example, do not necessarily need to be satisfied with the ruling: they may decide to appeal to the decision. In which case, the genleir chooses another set of verifications – except for this time, their number jumps to 11. The decision can also be appealed, which selects Genlair to another ruling and 23 verifications, then 47 verification, then 95, and so on and further.
The idea is to rely on the jury theorem of the condorset, which according to the Genlair’s pitch deck states that “When each participant is more than not a correct decision, the possibility of a correct majority result increases significantly because the group is bigger.” In other words, Jenler gets knowledge in the crowd. More verification involves, more likely they are zero on an accurate answer.
“This means that we can start small and very efficiently, but also that we can move forward at a point where something is very, very difficult, they can still be right,” Castelana said.
Castilna said that it takes about 100 seconds to process the average transaction, and the court’s decision is final after 30 minutes – a time limit that may be longer when many appeals. But this means that the protocol can reach a decision on major issues at very short time, day or night, rather than that it may take to go through the real real -world litigation processes which may take even in months or years.
In view of encouragement
Genlair’s mission naturally raises a question: is it possible for the system to game? For example, what if all verification selects the same AI (called Chatgpt) to solve a given proposal? Would this not mean that CHATGPT would have essentially issued the ruling?
Every time you query an LLM, you produce a new seed, Castelana said, so you get a different answer. At the top of it, verifications have the freedom to choose which LLM is to use on the basis of the subject. If this is a relatively easy question, there is probably no need to use an expensive LLM; On the other hand, if the question is particularly complex, the verification can choose for a high quality AI model.
Verifications may also end in a situation where they feel that they have seen a certain type of question, so they can pre-educate a small model for a specific purpose. “We think, over time, just going to be endless new models,” Castelana said.
The victorious side of the decision-making process has a strong incentive for verifications, as they are economically rewarded for it-while the side of the necklace ends the costs associated with using computation, assembled without any prize.
In other words, the question is not whether someone’s verification is providing a correct answer, but is it on the side with majority.
Since the verifications have no idea what other verifications are voting, the goal is to use the resources required for them to provide accurate information with the expectation that other verifications will also converge that information – because it will require possibly strict coordination to reach the same wrong answer.
And if he does not work the gambit, then the appeal is ready to kick in the system.
“If I know if I am reusing a good LLM, and I think other people are using a bad LLM and that’s why I have lost, I have a great encouragement to appeal, because I know that with more people, with more people, there is going to be an encouragement for them to use better LLM” because other recognized people want to earn a successful appeal.
The system makes it difficult to verify to verify, as they have only 100 seconds to reach a decision, and they do not know if they will be chosen to deal with specific questions. Castilna said that a unit would need to control between 33% and 50% networks.
Like the Ethereum, Genlayer will use a native token for its financial encouragement. With the launch of the testnet already, the project should be live by the end of the year, according to Castailna. “There is going to be a huge encouragement for people to come and build things on top,” he said.
