Let AI Agents Ship Your Freight

Point an agent at Warp and let it ship

You already run agents for code, support, and finance. This page is about pointing one at freight. Give it your Warp API key, drop in the endpoint reference, and the agent gets every shipping action as a callable tool: price a lane, place the order, pull a tracking update, fetch the BOL. No screen scraping. No EDI mapping layer to maintain.

The reason this works on Warp and not on a legacy portal is the surface. Older freight stacks hand you a login screen, a phone queue for rates, and EDI batches that need a custom parser per trading partner. An agent cannot click a button or wait on a fax. Warp exposes the same network as plain HTTP plus a typed CLI, so the agent reads a schema once and operates from there — closer to a freight control tower than a dashboard with buttons.

Wire it up with whatever you already use: Claude Code from the terminal, a LangChain or CrewAI tool, or a plain Python function calling the REST API. The agent runs the loop; you set the spend caps and mode rules. That is the on-ramp to fully managed freight.

What you can hand the agent

Each card below is a job you delegate, not a concept. Wire the tools, set the rules, and the agent owns it from your order system forward.

Integration patterns

$ claude "Ship 8 pallets of electronics from 90210 to 10001.
         Find the cheapest option with 3 day transit or less."

Claude Code: Querying Warp API for rates...

  Mode    Rate      Transit
  LTL     $1,247    3 days    ← Recommended
  Box     $1,890    2 days
  FTL     $2,340    2 days

Booking LTL option...
Booked: shp_01HGA2K9... | Pickup: April 7 | ETA: April 10

# Pseudocode — use REST API calls in production
import requests

API_KEY = "warp_sk_..."
BASE = "https://api.wearewarp.com/api/v2"
headers = {"x-api-key": API_KEY, "Content-Type": "application/json"}

# Quote
quote_res = requests.post(f"{BASE}/freights/quote", headers=headers, json={
    "origin": {"zipCode": "90210"},
    "destination": {"zipCode": "10001"},
    "items": [{"pallets": 8, "weight_lbs": 4800}]
})
quotes = quote_res.json()["options"]

# Select cheapest within transit window
best = min(
    [q for q in quotes if q["transit_days"] <= 3],
    key=lambda q: q["rate"]
)

# Book
book_res = requests.post(f"{BASE}/freights/booking", headers=headers, json={
    "quote_id": best["id"]
})
shipment = book_res.json()
print(f"Booked: {shipment['id']} | ETA: {shipment['eta']}")

@app.post("/warp-webhook")
async def handle_event(event: dict):
    if event["type"] == "shipment.exception":
        # AI agent decides what to do
        await agent.handle_exception(event)
    elif event["type"] == "shipment.delivered":
        await agent.reconcile(event["shipment_id"])

What you get when you point an agent at Warp

• -One key exposes the whole shipping surface — quote, book, track, invoice, documents — as callable tools, not a portal login.
• -Replies come back as typed JSON the agent reads directly. Nothing to scrape, no PDF to OCR, no per-partner EDI parser.
• -Bookings are idempotent, so a retried tool call never double-ships an order.
• -Status arrives by webhook push, so the agent reacts to a missed scan the moment it happens instead of polling on a timer.
• -A typed CLI mirrors every endpoint, so a terminal agent like Claude Code works with shell commands out of the box.
• -A test mode walks the full quote-to-delivery path with no carrier dispatched, so you harden the agent before it spends a dollar.

The network behind the API

• -24,000+ FTL carriers dispatched through the Warp driver app.
• -14,000+ box trucks and cargo vans.
• -50+ Warp operated cross-dock facilities across 12 metros.
• -Our AI backbone, Orbit, monitors every shipment for exceptions in real time.
• -All inclusive pricing. No fuel surcharges, no hidden fees.

Frequently asked questions