Prom (@prom_io): PROM: June & July Recap

PROM: June & July Recap April and May left PROM with a layered model on paper: the Agent Interaction Layer, the Verification Network, the Settlement Rail, and the PROM Network underneath them. The two months that followed were spent building outward from that structure rather than rearranging it — fourteen new ecosystem collaborations, continued development across all three layers, and one more article working through the part of the stack that decides whether an agent's output should be accepted at all. Strategy The thesis has not moved: AI agents are becoming economic participants, and participation requires more than execution. What has moved is the composition of the ecosystem forming around that idea. Earlier partnerships leaned toward AI services and general agent tooling. June and July leaned toward the harder edges of the stack — proof, risk, and financial flow. That shift matters more than the count. An economic layer is only as useful as the signals it can act on, and most of this period's additions supply exactly those signals: whether work was really performed, whether a counterparty can be trusted, whether a condition has actually been met. PROM's role in each case stays the same and stays narrow. Partners establish facts. PROM decides what happens to value once those facts are established. Development Active work continues across all three components of the PROM AI stack: the Agent Interaction Layer — where agents identify themselves, expose capabilities, and exchange requests; the Verification Network — where outcomes are evaluated, attested, and challenged; the Settlement Rail — where verified outcomes turn into escrow, conditional release, and payment. Each is being developed further in parallel, and the pieces are increasingly built with reference to one another rather than in isolation. Partnerships Fourteen collaborations were announced between June and July. Grouped by what they contribute, rather than by date: Proof and verification. Origins Network was the strongest fit of the period — modular infrastructure for verifiable computation and trust-minimized execution, which is the exact input the Settlement Rail is designed to consume. ExpandZK approaches the same problem from the data side, authenticating real-world and Web2 information without exposing what should stay private. In both cases the connection is direct: value should move on a proof, not on an assertion. Agent commerce and orchestration. TermiX is building an agent commerce layer where autonomous systems discover work, coordinate execution, verify outcomes, and settle — a full lifecycle that overlaps the PROM stack layer for layer. LinkLayerAI contributes verifiable trading agents and orchestration between agents and external services. XAgent lowers the barrier further up, letting agents be created and deployed in natural language. Together they describe the demand side: more agents doing more work that eventually needs to be paid for. Trust and risk. ChainAware brings wallet intelligence and behavioral analysis; Aegis AI brings threat detection and risk assessment ahead of execution. Neither is an economic layer, and neither needs to be. They answer a question that has to be answered before value moves — who can be trusted, and is this safe to execute — and PROM connects that answer to what happens next. Financial flows. Haven AI applies autonomous execution to capital allocation across on-chain strategies and real-world assets. Wager Predict prices future outcomes through permissionless prediction markets. Both produce conditions that settlement logic can act on: a strategy completes, an outcome resolves, a payment follows. Content and consumer surfaces. AIVIVE ties AI-driven creation to on-chain economics, so that usage and value stay connected rather than drifting apart. ArcNova explores the same question for AI-generated entertainment and storytelling. Magne AI works closer to the user, building a Web3-native device where secure access and native payments sit in the same place. Also joined the ecosystem. HPX and Blazpay, expanding the surface around wallets, payments, and AI-assisted DeFi interfaces. Long-form Content Part V, Building the Verification Layer, continued the series on the AI-agent economy. Where earlier parts described how agents identify themselves and interact, this one addressed the question that follows: once an output exists, who decides whether it should be accepted? The article makes a distinction the rest of the stack depends on — validators act as arbitrators, not as second executors. They do not redo the work; they rule on it. It also looks at what agent verification can borrow from existing dispute systems such as UMA and Polymarket, and how a verification decision becomes economically final through PROM. Verification produces decisions. PROM makes them binding. Conclusion June and July added density rather than direction. The layered model stayed where April and May left it, development continued across all three components, and the ecosystem around the stack grew by fourteen collaborations weighted toward proof, trust, and financial flow. The gap that remains is the same one as before, and it is a practical one: an architecture with partners attached is still not an architecture in use. The next stage is turning these connections into working flows — agents that discover, execute, prove, and settle in a single path, rather than four capabilities that fit together on a diagram. More updates are coming as we continue building.