12.2 · PoB and Agent collaboration
From manual splicing to system operation: standards, logs, permissions and replicable methods are aligned at the same stage.
The first stage proves that “the network can run”; this stage proves that “the network can run on a large scale” - PoB becomes a composable trust ledger, and the Agent becomes a constrained execution layer rather than a black box automation.
Industry reference
- Uniswap v2→v3: From a general pool to centralized liquidity and fee tiers, it is a dual evolution of "rule refinement + capital efficiency"; WCN Phase 2 corresponds to the versioning of PoB fields, audit levels, incentives and attribution rules.
- Ethereum L2 route: Layer execution and settlement, and the main network ensures finality; WCN layers "human decision-making/Agent execution/PoB records", and Agent does not replace compliance and final review boundaries.
- Mature Web2 Market: Uber/Airbnb complete the middle layer of trust and security before scaling; Phase 2 completes auditable automation rather than simply adding the number of functions.
specific deliverables
PoB stabilizationDelivery: PoB schema version, change log, review SLA, objection and reconsideration process; PoB template library by Deal type. Indicators: Cross-month PoB structure consistency, review backlog days, attribution dispute rate trend.
Agent formalizationDelivery: Role permission matrix, tool whitelist, session/task log retention policy, human-in-the-loop trigger conditions. Indicators: Agent output adoption rate, number of unauthorized interceptions, number of rollbacks and patches (stability agent).
Node methodology productizationDelivery: The playbook (checklist, templates, training materials) of high-value nodes enters the system. Indicators: The time it takes for a new node to reach "first order qualification", playbook usage rate and output comparison.
closed loop densityDelivery: The number of Deals that can be advanced in parallel on the same network within unit time, and resource conflict rules. Indicators: Deal throughput per person/node, stage stagnation rate (which card has the most).
Success Metrics (Example)
- PoB v1.x During the freezing period, breaking changes will only be released through governance/version numbers, and there will be a migration guide.
- Agent-related incidents (wrong referrals, data leaks, unauthorized operations) have postmortem and rule patches that can be restored.
- Replicable: New region or new vertical achieves funnel metrics (as defined by operations) comparable to baseline within 90 days under a preset playbook.
Signals for entering the third stage (summary)
Settlement objects (who pays to whom, which PoB is based on, how disputes are resolved) can be aligned on paper and in the system; the identity and authority model is sufficient to support accounting and auditing without relying on temporary tables. See [12.4](/wiki/roadmap/12-4-Milestones and Phase Switching Signals) for details.
Main risks at this stage
Automated trust mismatchAgent speeds up error propagation. Release: hierarchical automation, shadow mode, random inspection and red teaming.
PoB over-engineeringToo detailed standards slow down business. Release: Take the "minimum auditable set" as the bottom line, and the remaining fields are gradual.
Organizations rely on hero nodesScaling still relies on a few people. Release: Mandatory documentation and handover mechanisms are included in KPIs.
Capitalize in advanceNFT/Token is out of touch with the rhythm of business. Release: The asset layer will only initiate projects after the 12.4 threshold is met.
Why this phase must come after Phase 1
Without stable Deal and Proof samples, PoB standards will become a castle in the air; without clear manual processes, Agent will only amplify chaos. This is the same as Rollup becoming a rigid requirement after L1 fees and congestion "actually occur" - the complexity should be driven by the real load.
At the end of the second phase, it should be able to demonstrate to the outside world how human-machine collaboration can stably produce more verifiable results under the same set of rules, and that any incidents that occur can be held accountable and patched.