What personal markets still need to prove?
The next useful step is not to draw an exchange frontend and then find a way to call it decentralized. It is to test whether independent personal infrastructure can improve discovery and negotiation while keeping authorization under the user's control.
This claim can fail. If every agent depends on the same indexer, model provider, solver group and order gateway, we may merely have hidden concentration behind more components.
Start with a bounded experiment
A small crypto-to-crypto OTC community would provide a tractable setting. Limit assets, amounts and deadlines. Require quotes to specify fees, received amounts and failure conditions. Compare manual inquiry, a single aggregator and personal agents on executable outcomes, total cost, delay, privacy exposure and human intervention.
Many automated inquiries are not evidence of value. If the agent spends money rediscovering public prices, it may be worse than a simple router. If it improves outcomes through waiting, splitting or finding complementary needs, identify which step produced the benefit and when it stops working.
A DEX as a Blocklet, or a narrower Token Vending Machine Blocklet, could be an experimental delivery form. A Blocklet is an application module that can be deployed and run. A vending-machine design would expose an owner's inventory and explicit rules. These are research directions, not product delivery commitments or an announcement that ArcBlock will operate the venue.
Evidence we would need
| Question | Useful evidence |
|---|---|
| Can offers remain reachable? | Replication, expiry, withdrawal and recovery tests |
| Can discovery providers be replaced? | Comparable results and migration across indexes |
| Are quotes executable? | Concurrency, balance-change and cancellation failure rates |
| Does the agent represent the user? | Incentive disclosure and adversarial authorization tests |
| Do settlement semantics hold? | Chain-specific signature, nonce and asset-behavior tests |
| Do credentials preserve privacy? | Freshness, revocation and correlation analysis |
| Who has operating duties? | Review of actual control, customers, fees and jurisdictions |
ArcBlock's exchange primitive needs explicit lifecycle behavior for any new open-order design. Ethereum and Solana adapters should reuse mature security practices. Cross-chain mechanisms need their own threat model rather than inheriting same-chain atomicity language.
Keep the useful intermediaries
Market makers can supply inventory, brokers research, OTC desks specialized networks, solvers better paths, and CEXs efficient service. The objective is not to make everyone maintain a global order book. It is to make participation possible without automatically surrendering context and execution authority to one operator.
Some frequent property investors obtain their own agent license because they want to represent their interests more directly. A personal AI agent is not a legal license, but the aspiration is similar: own a tool that works for you.
White-label software lets more people run exchanges. A personal market node could let people own their terminal and representative, even without a public website. ArcBlock's chain primitives, ARC, DID and personal-space architecture are interesting because independent participants might combine them in this way.
Whether that is economical remains an experimental question. The principle is clear enough to guide the experiment: intermediaries should earn their place through the value they provide, while users retain the ability to choose them.