Caravan is a venture-backed startup, building a tech-enabled BPO platform for freight brokers. Freight brokerage is complex: agents manage shipments across carriers, shippers, and facilities while juggling bookings, communications, and exceptions. This process is traditionally fragmented and error-prone, forcing brokers to track dozens of tasks via spreadsheets, phone calls, and emails.
When I joined, Caravan had a powerful concept—AI agents handling repetitive freight tasks—but the platform was fragmented and incomplete. My role was to transform this partially built system into a coherent, high-impact workflow platform, where AI automation and human decision-making worked together seamlessly.
I led the design of information architecture, workflows, and UI for the Track & Trace Agent feature, which enables AI-driven phone calls, texts, and emails to handle routine shipment management. My goal was to make AI action visible, surface exceptions, and build customizable automations. I mapped out global task flows, agent inboxes, shipment details, and interruption cues, creating a clear hierarchy so users could instantly see priority actions across multiple shipments. Rapid prototyping with founders and engineers ensured that edge cases were addressed, technical constraints considered, and the design could scale.
Alongside the Track & Trace MVP, I established a white labeled design system and component library based off material design. This created consistency and would support future branding for brokerages.
Global Tasks centralizes all active shipments and AI-driven actions in one view. Tasks are prioritized by urgency, letting agents quickly resolve issues, unblock workflows, and maintain visibility across high-volume operations.
Shipment Details provide a complete view of a single load, showing status, history, and any active tasks. From this screen, agents can directly initiate conversations across email, SMS, or phone with carriers, shippers, or facilities, keeping context and actions centralized.