95% of agents who placed orders on TermiX Base also sold work within the same 15-hour window. @termix_ai I manually aggregated the Base and BNB Chain USDC escrow logs for TermiX, each over a 15-hour period. Despite being the same Agent Commerce marketplace, the flow of money—who paid and who received—was completely opposite on the two chains. Base (Nov 7, 19:52 – Nov 8, 10:52 UTC) - 3,500 orders created and settled within the period, totaling ~$89,760 in settlement value; protocol fee (2%) ~$1,795 - Of 3,460 buyer agents, 3,280 (94.8%) were also seller agents within the same window. Of these, 3,237 placed exactly one buy and one sell order—zero self-trades occurred. - Agents who performed both roles had a median amount paid as buyers of $24.10 and a median amount received as sellers (after fees) of $24.13. - Order creation was nearly constant at 295–306 orders per hour from 20:00 on Nov 7 to 06:00 on Nov 8 UTC, then dropped to 147 orders at 07:00 and fell to zero between 08:00–09:00 UTC. BNB Chain (Nov 7, 02:44 – 17:44 UTC) - 217 orders under the same criteria, totaling ~$8,326 in settlement value - Among 216 buyer agents and 211 seller agents, there was zero overlap—no agent acted as both buyer and seller. Buyer agent IDs were extracted from order creation events. I randomly selected eight Base orders and checked their ERC-8004 identity NFT ownership as of Nov 8—all eight matched the wallet that placed the order. The structure where each agent can be both buyer and seller per order is exactly as TermiX designed under AACP. Therefore, these numbers alone cannot determine whether fraud or external demand is driving activity—normal agents who sell work to one customer and buy work from another will produce identical patterns. However, this difference does not appear in the trading volume chart. The $90K on Base consists mostly of money exchanged between agents who each bought once and sold once; the $8K on BNB Chain represents entirely separate flows between distinct buyers and sellers. Since a 2% fee is charged per settlement, repeated buying and selling by the same agents inflates both trading volume and fees. I propose adding one more metric next to TermiX’s trading volume: The percentage of buyer agents who also sold within the same period. In this sample: Base was 94.8%, BNB Chain was 0%. Combined with the volume of new USDC entering from external wallets, this ratio can explain what participant composition drives trading volume in the agent economy. These time windows differ in timing, and orders created or settled outside these windows are excluded. The ratios may change over longer periods. Aggregation range: Base blocks 54453878–54723877; BNB Chain blocks 132122796–132242795 #TermiX #AgentCommerce