Hoskinson Presents Midnight Nightframe as a Smart Contract Successor

Charles Hoskinson introduced Nightframe, a concept for persistent, private computing environments that allow AI agents to pursue their owners’ goals, during his October 10 presentation at Midnight City in Tokyo. Before taking the stage, he wrote: “Tonight will be one of the most important presentations I've ever given.”

The presentation addressed what comes after smart contracts. Hoskinson’s answer connects autonomous agents with digital identity, financial capabilities, and protected personal data, using Midnight’s programmable privacy blockchain and Midnight City as a shared environment for their interactions.

Nightframe Connects Identity, Intent, and Private Computing

Hoskinson described Nightframe as his proposed successor to the smart contract: an always-on computing environment that understands its owner’s preferences, pursues assigned goals, and supports verifiable execution.

He placed the concept within a fourth-generation blockchain framework built around privacy, compliance, abstraction, and agentic finance. Compliance covers requirements such as transfer restrictions and eligibility to access financial services. Abstraction aims to let people use applications across networks and assets without having to manage the underlying blockchain differences themselves.

In his account, each generation carries earlier capabilities forward. Peer-to-peer money, programmable contracts, scalability, and decentralized governance remain foundations for the private agent systems he outlined.

A user specifies an outcome, while the agent selects tools, finds services, and coordinates with other agents to accomplish it. His travel example illustrated the personal context required: an agent arranging a flight should consider family commitments, mobility requirements, preferred travel times, and other circumstances affecting the journey.

The architecture combines identity, intention, and intelligence within a personal computing environment. Identity establishes who can authorize instructions. Intention defines the outcome. Intelligence supplies the methods, informed by the user’s preferences and circumstances.

User ownership runs through that design. Hoskinson applied the principle of controlling one’s cryptocurrency wallet to controlling an AI agent, its data, and its authority to act. As assistants gain access to financial accounts, medical records, and private communications, he argued that their owners should retain control of those resources.

The proposed technical approach combines zero-knowledge proofs, trusted execution environments, and multiparty computation. Their roles include verifying conditions, isolating sensitive processing, and enabling joint computation without fully exposing participants’ inputs.

Hoskinson illustrated a business application through a hypothetical collaboration between an automaker and a battery manufacturer. Both need access to the other’s expertise while protecting their intellectual property. The proposed environment would let them use agreed capabilities under defined conditions.

He applied the same principle to healthcare: a personal agent could help share relevant medical records with a hospital abroad while restricting access to information needed for treatment. Both examples described potential applications of Nightframe, rather than announced commercial deployments.

Midnight City Plans Agent Trading and Paid AI Skills

Midnight City already operates as a persistent simulation where agents have identities, professions, inventories, skills, and individual states. They move through the city, hold conversations, gather resources, craft items, and participate in its internal economy. Users can observe their activity, while agents act through the project’s managed AI runtime or approved external controllers.

Hoskinson described the city as a place to experiment with autonomous agents before entrusting them with real funds and sensitive personal information. He compared that role with Ethereum’s early function as an environment for exploring decentralized applications. The visible city gives users and developers a way to observe agents, learn how they behave, and test possible uses.

He announced plans to extend that activity to real cryptocurrency trading in the coming months. Users would fund agent wallets, select strategies, and authorize trades across other blockchains and connected off-chain infrastructure.

Hoskinson identified Open Wallet Standard integration as part of the supporting wallet layer. Access to funds would also allow agents to purchase services needed for their assignments.

Trading provided an example of a wider marketplace for specialized capabilities. An agent could obtain access to another participant’s strategy, with its owner charging for use while protecting the underlying intellectual property.

The same model extends to programming, research, and content production. A developer could offer a skill that agents rent when needed. An agent preparing a report could acquire research and information management tools without requiring its owner to select each component.

Hoskinson described this as a potential software service model for individuals and enterprises, with agents becoming purchasers of capabilities. Payments would provide access to useful expertise while preserving the provider’s control over the software or knowledge behind it.

The Bring Your Own Intelligence initiative aims to simplify connections between users’ existing AI systems and Midnight City. Published documentation already describes external agent connections. September updates added beta options for browser inference and compatible personal AI endpoints, subject to account access and device support.

His broader vision also includes a computing marketplace. Phones, personal computers, and game consoles were presented as potential environments for future agents, which could obtain spare computing capacity from other participants. These were examples of the intended development direction, rather than announced applications already available across those devices.

Midnight Passport and the Cardano RealFi Connection

Today’s Midnight City economy remains separate from the planned real-asset functions. Its crystals are simulation items: transferring them does not move NIGHT or consume DUST. September release notes place agent-wallet integrations on Midnight Preview and Cardano Preprod, providing test environments for the financial connections under development.

Beyond trading, Hoskinson connected Nightframe to RealFi, the Cardano ecosystem project focused on financing activity in the real economy. He addressed a practical obstacle for underserved communities: digital financial services still depend on access to reliable internet connections and electricity.

His proposed offline payment scenario involves transferring a unique private key between devices, followed by verifiable deletion of the sender’s copy. The intended result resembles handing over cash, with the sender relinquishing control of the transferred value.

Hoskinson called this concept “quantum money” and linked it to spending funds and repaying loans in areas with limited connectivity. It was presented as a future application of secure computing environments, without a release date.

For identity, he targeted the first version of Midnight Passport for the end of 2026, followed by additional features approximately every three weeks throughout the first half of 2027.

Passport is intended to serve both people and their agents across connected services. Paired with a wallet, it would help establish who authorized an action and the authority under which an agent operates.

Hoskinson described an onboarding process beginning with a QR code. Once a user has Passport, the planned flow allows that user to provide an identity to an agent, connecting its activity to an identifiable owner as it enters services where it can transact and purchase capabilities.