Supply Chain and Logistics

22/810 → 12/622; -0.79 percentage points of formation share.

What counts as "supply chain and logistics" in this batch has narrowed sharply from moving goods to automating the paperwork and phone calls around moving goods: Trava audits customs entries, Comena and Kanava AI replace manual order-taking and inbound calls, and Hemut and Lunavo target the back-office and dispatch layer of trucking rather than the trucks themselves. Even the load-planning entrants, Maximal and Fleetline, are converging on the same dispatcher's desk from opposite starting points, suggesting the actual hard problem left in freight is context-aware scheduling rather than tracking or booking. Physical-world sensing shows up only at the margins, with Rebulk applying LiDAR and computer vision to bulk commodities that never got barcodes in the first place. The formation's cooling to 0.71x share change reads oddly against Manufacturing accelerating at 3.24x, since these agents are largely built to sit on top of manufacturers' and distributors' existing order flow, implying the money is following production upstream faster than it's following the logistics layer that depends on it. None of this required new sensors or new freight infrastructure to become possible, just the 2023-24 jump in voice and document-parsing models good enough to handle messy customs filings and call center chatter, which is why a wave of these agent-layer companies all cluster in 2025 rather than earlier.

No comparable outside activity is connected yet.