The following is a guest position and an opinion of Matej Janež, responsible for partnerships in Oasis.
At Ethdenver earlier this year, a subject kept coming again and again: IA and Autonomous AI agents. This excitement took place in other cryptographic conferences during the year. There are also good reasons for excitement: they are no longer just ideas – they are there, and they manage real funds – but their dependence on transparent blockchains can become their greatest weakness.
What are these AI agents exactly? These are smart software that works alone to manage specific tasks. In Crypto, they can use automatic learning and blockchains to monitor markets, spot patterns and automatically manufacture trades. Unlike traditional commercial robots, today AI agents are adaptive; They refine their continuous behavior depending on what gives the results.
But there is a big problem that has been underestimated and misunderstood: the fact that these onchain agents work on transparent blockchains make their decision -making – their “brains” – essentially public. This opening creates real obstacles for agents who try to compete in the financial markets.
AI agents in Defi
Currently, agents manage trade in decentralized exchanges, manage loans and optimize return agriculture. They instantly react to market changes, often making rapid decisions with a lot of money. Clever. Fast. Effective.
But they are faced with a basic challenge. The very system that allows them to work – public blockchains – shows their strategies for everyone. Each transaction, each interaction with an intelligent contract, leaves a path that reveals how they “think”. It is not very different to play poker with your cards facing the table.
Of course, we could manage these strategies on private servers and only submit final transactions to the blockchain, but this is fundamentally against the objective of the promise of crypto transparency and the verifiability of onchain. The full point of DEFI is to remove the need for trusted third party and centralized systems.
Consider what is already going on in Defi today. A performance bot of yield agriculture is continuously scoring the best yields between protocols, moving millions between loan platforms according to subtle market changes. If his strategy becomes visible on the channel, the competitors simply look at the pools he enters and leaves, which thresholds and with what timing – then cloning the strategy without the research costs. In the decentralized credit markets, AI agents that mark portfolios for subculturalized loans become useless if borrowers can see exactly what behaviors improve their scores, leading to artificial wallet models designed to play the system.
The most concerning could be Dao Treasury agents – When their rebalancing strategy is transparent, everyone can make major liquidity movements, effectively flying the community with each transaction. These are not on -board cases; These are fundamental defects in the application of AI to transparent systems where the execution of the strategy and the development of the strategy are impossible to separate.
Without doubt, the worst of all is the potential for manipulation of the market. When bad players understand how an agent makes decisions, he can create situations designed to deceive him. Transparent agents are easy targets.
Why a “private brain”?
A “private brain” for deficiencies to solve these problems. By keeping confidential calculations, the agents could make decisions without showing their logic or their intentions until transactions pass.
The security benefits are obvious. The strategies remain protected against copies. The first cycle becomes more difficult without seeing the transactions pending. The work of the agent remains private. Teams that build better algorithms can keep their advantage, creating reasons to continue to improve. The market rewards real improvement instead of copying quickly. On the larger scale, the markets would become more stable. When agent strategies remain secret, you avoid breeding – where several agents follow identical strategies. This reduces market correlated movements and reduces risk -scale risk.
If we continue as we are now – if the deficiencies continue to operate with glass box brains, we must worry about a few things that happen.
Market exploits will become more common and sophisticated. As the agents manage more funds, the awards to exploit them also increase. Without confidentiality measures, these exploits become simple technical exercises rather than difficult security violations.
The cannibalization of strategy is just as disturbing. When winning strategies are copied quickly, they also stop working. Finally, all agents use similar approaches, creating monoculture. The market loses variety and resilience.
This leads to what you could all the problem of “the spirit of the hive”; When all agents operate in the same way, they will also react to market changes in the same way. This makes market oscillations larger, increases volatility and creates the risk of flash planting when the conditions trigger the identical responses widespread. What starts as individual agents becomes, fundamentally, a massive entity with system -scale effects. To spell it: these are not the ingredients of a healthy market.
Technical solutions
Confidence execution environments (Tees) offer a solid way to create these private brains. TTT-SOUCT Office where the calculation occurs in isolation, even protected from the system hosting it. You can check that the work has occurred properly, but the details remain private.
This technology allows us to balance openness and privacy. The framework of an agent can be public and verifiable, while decision -making and specific strategy details remain protected.
The addition of private calculation to deficits is not only useful – it is necessary that algorithmic finance is developing properly. Without intimacy, we build a market where innovation is punished, the exploitation is rewarded and the system risks accumulate below the surface.
We are at a critical funding of funding fed by AI where our choices will determine whether autonomous agents create a more efficient or dangerously fragile market. Private calculation technology exists today, but implementation requires deliberate action by manufacturers and protocols. As financial intelligence moves more and more on the chain, ensuring that these systems can work with calculation confidentiality will not only protect individual strategies – it will save the integrity of the entire DEFI ecosystem.
