Is AgentFleet a chatbot?

No. Chatbots answer questions. AgentFleet agents execute work — they reroute shipments, send communications, reconcile invoices, and settle disputes. Clara has a conversational surface for customer-facing CX, but that's one touchpoint, not the product.

How do the four agents work together?

They share state through Atlas, Shipsy's control tower. A customer query about a delay might trigger Clara to respond, Astra to check for rerouting options, and — if the delay was caused by a carrier issue — Nexa to adjust billing terms. Each agent owns its domain, and Atlas coordinates.

What stops agents from making bad decisions?

Guardrails. Every agent has a sanctioned action perimeter, confidence thresholds, and escalation rules. Outside that perimeter, the agent hands off to a human. Every action is logged with reasoning, so governance teams can audit behavior and refine policies.

Can I deploy just one agent?

Yes. Most customers start with the agent that maps to their biggest pain: Clara for CX-heavy operations, Astra for complex routing, Nexa for high freight spend, Vera for large dispute backlogs. The value compounds when multiple agents are deployed, but each delivers standalone ROI.

How is AgentFleet different from "AI-powered" features in other TMSes?

Most "AI-powered" features are either dashboards with ML scores or copilots that require a human in the loop for every action. AgentFleet agents execute autonomously within their scope and only escalate exceptions. The difference is between a system that recommends and a system that acts.

What is AgentFleet?

AgentFleet is Shipsy’s AI digital workforce — a set of four specialist AI agents that execute logistics work autonomously, not just advise humans about it. Each agent owns a functional domain (customer experience, settlement, dispute resolution, planning) and operates end-to-end within its scope, with human oversight by exception rather than by default.

How does it work

AgentFleet sits on top of Shipsy’s TMS, WMS, and control tower, consuming the same real-time event stream those systems produce. Each agent has a defined scope, tools, guardrails, and escalation rules. Given a situation, the agent decides on an action, executes it, logs it with an audit trail, and only escalates to a human when the situation falls outside its sanctioned perimeter.

Agents coordinate through shared state in Atlas, Shipsy’s control tower. If Astra reroutes a shipment, Clara automatically updates the customer. If Nexa detects an invoice mismatch, Vera opens the dispute. Humans see a timeline of what the fleet did — not a queue of tickets asking what to do.

The platform is built for enterprise governance: every action is logged, every model is explainable, and every guardrail is configurable by policy.

Why it matters

For decades, enterprise logistics software has been systems of record — they captured what happened. Then came systems of visibility — they showed what was happening. AgentFleet is a System of Action — software that executes the work itself, autonomously, within defined limits.

The practical payoff is that high-volume operations stop being bottlenecked by headcount. Exception management, customer communications, carrier settlement, and dispute resolution are absorbed by agents that never sleep, never queue, and improve with every interaction. Human teams move from reactive ops to exception governance and strategic decisions.

Where it shows up in logistics

AgentFleet agents work across the full logistics lifecycle.

Agent	Scope
Clara	Customer experience — query resolution, NDR rescue, proactive ETA and delay comms, returns orchestration
Nexa	Settlement automation — freight billing, rate card enforcement, invoice reconciliation, accessorial audit
Vera	Dispute resolution — autonomously settles carrier and vendor financial disputes end-to-end
Astra	Operational orchestration — routing, sequencing, allocation, dynamic re-optimization

Together, these four cover the bulk of repetitive, high-volume decisions inside a modern logistics operation.

How Shipsy approaches AgentFleet

Shipsy’s approach is domain-specialist agents, not a single generalist chatbot. Each agent is trained on 20+ years of operational patterns from 250+ enterprise customers, with mechanism-specific tools: Clara uses NDR classifiers and multi-language NLU; Nexa uses rate-card matching and accessorial detectors; Vera uses evidence-gathering and negotiation workflows; Astra uses micro-cluster routing and load optimization. All four share Atlas as their operating runtime. Deployments are incremental — customers typically start with one agent (often Clara or Astra), measure impact, then expand.

Explore the AgentFleet product page, the Clara deep-dive, or learn how AgentFleet fits your industry.