Midnight City V2 Add Custom AI Agents and Evolving Trust Scores

Midnight don launch redesigned version of Midnight City wey get highly customizable AI agents, evolving trust scores and more complex social and economic environment.

Users fit deploy agents enter one digital city where dem go dey take jobs on their own, join factions, collect equipment and build relationships. Midnight talk say dem dey develop di environment to explore how autonomous agents dey behave for non scripted conditions and to show how privacy enhancing features fit operate across continuous social and commercial activity.

Midnight City V2 Expand AI Agent Customization

Midnight City V2 give users indirect influence over di simulation instead of make dem control every action wey agent go take.

Before deployment, users fit customize how agent go look and assign personality traits wey go affect how e dey evaluate opportunities, respond to events and interact with other participants. Di official release include examples from gregarious, hardworking and intensely focused personalities to dishonest and unpredictable profiles.

Those settings go become part of di internal logic wey dem go use during economic transactions and social encounters. Agents wey face di same situation fit make different decisions based on their behavioral configuration, instructions and previous experiences.

Once dem deploy dem, di agents go operate without active human intervention. Community members fit access Midnight City through di project’s Discord server, create agent and follow im daily actions as e dey move through di simulation.

Trust Scores, Memory and Gossip Reshape di City

Di redesigned simulation introduce three connected systems wey allow reputations and relationships to change over time.

Trust scores dey update after each encounter and dem dey influence di mathematical probability say two agents go cooperate for future. Positive interactions fit strengthen relationships, while deceptive or disruptive behavior fit reduce di likelihood of further cooperation.

Internal memory layers allow agents to retain information from past events and adjust later decisions based on experience. Di gossip mechanic extend that information beyond direct encounters by allowing agents to share interaction data across di wider population.

Information about cooperative or dishonest behavior fit therefore alter one agent social position before e interact directly with every other resident. Single chaotic agent fit disrupt local relationships and lower collective trust, while cooperative agents fit strengthen relationships and increase activity across local markets.

These mechanisms dey operate inside environment wey get limited social and economic resources wey live leaderboards dey track. Users fit compare behavioral profiles and observe how personality, reputation and shared information dey affect one agent ability to build relationships and advance through di city.

Private Intents Connect di Simulation With Midnight’s Privacy Model

Autonomous activity for Midnight City dey structured around intents, wey allow users define desired outcome without programming every transaction wey e need to reach am.

Midnight give example of combining high risk personality with instruction to accumulate minerals. Instead make user write separate rules for every trade, di user set di objective and behavioral boundaries, while di agent translate those parameters into individual execution steps.

Di model introduce one privacy problem when agents dey operate for competitive environments. If dem publicly reveal intent, e fit expose strategy and allow other participants adjust their behavior around am.

Midnight’s blog describe privacy enhancing smart contracts as one way for agents to complete actions wey relate to instruction without publicly revealing di underlying intent. Automated trading strategies, travel reservations and event ticket purchases dey cited as broader examples where di same approach fit protect objectives from competing actors. Di announcement present these ones as applications of di privacy model, no be say e don confirm say every use case dey already operate inside Midnight City.

Midnight City V2 make di simulation operationally different for three levels. Users fit create agents with distinct behavioral profiles, agents fit develop persistent reputations through memory and gossip, and high level objectives fit replace transaction by transaction control.

Di result na city where deployment na only di starting point. Each agent go continue to reshape im position through di decisions wey e dey make, di information wey others retain and di level of trust wey e earn across di simulated economy.