LASF$260|SFLA$264|COLLA$366|COLCHI$193|NJMIA$288|COLSF$420|SFSAC$142|LADAL$398|LASD$156|COLMIA$303|SFSEA$235|COLDAL$208|LASLC$297|LAPHX$244|LALV$260|LAORL$437|LANJ$447|HARNJ$188|LACOL$365|CHINJ$235|DALMIA$266|SFPDX$231|COLPHX$244|NJORL$304|SFSD$208|COLORL$310|CHIMIA$295|COLDEN$275|LAMIA$420|LVLA$215|SATAUS$125|LASAC$195|LADEN$310|DALLA$385|SFPHX$280|LASEA$340|NJDAL$335|ORLMIA$145|ORLTPA$130|DALHOU$155|DALSAT$165|NJATL$270|MIANJ$305|NJCHI$240|NJLA$440|ORLJAX$140|COLSLC$320|HOUNJ$345|SLCBOI$185|LAPDX$315|LASF$260|SFLA$264|COLLA$366|COLCHI$193|NJMIA$288|COLSF$420|SFSAC$142|LADAL$398|LASD$156|COLMIA$303|SFSEA$235|COLDAL$208|LASLC$297|LAPHX$244|LALV$260|LAORL$437|LANJ$447|HARNJ$188|LACOL$365|CHINJ$235|DALMIA$266|SFPDX$231|COLPHX$244|NJORL$304|SFSD$208|COLORL$310|CHIMIA$295|COLDEN$275|LAMIA$420|LVLA$215|SATAUS$125|LASAC$195|LADEN$310|DALLA$385|SFPHX$280|LASEA$340|NJDAL$335|ORLMIA$145|ORLTPA$130|DALHOU$155|DALSAT$165|NJATL$270|MIANJ$305|NJCHI$240|NJLA$440|ORLJAX$140|COLSLC$320|HOUNJ$345|SLCBOI$185|LAPDX$315|View all rates →LASF$260|SFLA$264|COLLA$366|COLCHI$193|NJMIA$288|COLSF$420|SFSAC$142|LADAL$398|LASD$156|COLMIA$303|SFSEA$235|COLDAL$208|LASLC$297|LAPHX$244|LALV$260|LAORL$437|LANJ$447|HARNJ$188|LACOL$365|CHINJ$235|DALMIA$266|SFPDX$231|COLPHX$244|NJORL$304|SFSD$208|COLORL$310|CHIMIA$295|COLDEN$275|LAMIA$420|LVLA$215|SATAUS$125|LASAC$195|LADEN$310|DALLA$385|SFPHX$280|LASEA$340|NJDAL$335|ORLMIA$145|ORLTPA$130|DALHOU$155|DALSAT$165|NJATL$270|MIANJ$305|NJCHI$240|NJLA$440|ORLJAX$140|COLSLC$320|HOUNJ$345|SLCBOI$185|LAPDX$315|LASF$260|SFLA$264|COLLA$366|COLCHI$193|NJMIA$288|COLSF$420|SFSAC$142|LADAL$398|LASD$156|COLMIA$303|SFSEA$235|COLDAL$208|LASLC$297|LAPHX$244|LALV$260|LAORL$437|LANJ$447|HARNJ$188|LACOL$365|CHINJ$235|DALMIA$266|SFPDX$231|COLPHX$244|NJORL$304|SFSD$208|COLORL$310|CHIMIA$295|COLDEN$275|LAMIA$420|LVLA$215|SATAUS$125|LASAC$195|LADEN$310|DALLA$385|SFPHX$280|LASEA$340|NJDAL$335|ORLMIA$145|ORLTPA$130|DALHOU$155|DALSAT$165|NJATL$270|MIANJ$305|NJCHI$240|NJLA$440|ORLJAX$140|COLSLC$320|HOUNJ$345|SLCBOI$185|LAPDX$315|
WARP // FREIGHT NETWORK191,000+ ADDRESSES DELIVERED TO

AI agents

Best Freight API for AI Agents

AI agents that procure freight need APIs designed for machine consumption. Not web portals. Not PDF rate sheets. Not phone trees. Structured JSON, standard error codes, idempotent operations, real time webhooks, and documentation that an LLM can read in a single pass. Most freight platforms were built for human operators. This page evaluates which APIs are ready for autonomous agents.

5Calls from quote to POD
12+Webhook event types
JSONEvery endpoint, every response

What AI agents need from a freight API

An AI agent is software that acts autonomously to accomplish a goal. In freight, that goal is moving goods from origin to destination at the best price and service level. The agent needs to quote rates, compare options, book the best one, monitor the shipment, handle exceptions, and close out the order with invoice and proof of delivery.

To do this without human intervention, the agent needs an API that speaks its language. That means structured data (JSON, not HTML), explicit field names (not abbreviations or codes that require a lookup table), standard protocols (HTTP, not proprietary VAN connections), and event delivery (webhooks, not polling or phone calls).

Agent readiness: Warp vs traditional freight platforms

CriteriaWhat agents needWarpTraditional freight
Response formatStructured JSON with typed fieldsJSON with explicit field names and enum valuesHTML portals, PDF rate sheets, phone calls
Error handlingStandard HTTP codes + structured error body400/401/422/429/500 with { error: { code, message, details[] } }Generic error pages, no structured error data
IdempotencySafe retries without duplicate bookingsQuote IDs prevent duplicate bookings. GET operations are inherently idempotent.No idempotency guarantees. Retries may create duplicates.
Event deliveryWebhooks (push), not polling12+ webhook event types pushed in real timeCheck call updates via phone or email
DocumentationMachine readable, parseable by LLMsStructured text reference, OpenAPI spec, llms-full.txtPDF manuals, portal help pages
CLICommand line interface with parseable outputYes. Quote, book, track from terminal.No CLI. Web portal only.

Agent workflow: Quote to proof of delivery

An AI agent completes the full freight lifecycle in five API calls. Each step returns the identifiers needed for the next call. No human intervention required at any step.

1
POST quote

Get rates

Agent sends origin, destination, items. Gets quoteId and totalCost.

2
EVAL Business logic

Select best option

Agent applies rules: cost thresholds, transit requirements, carrier preferences.

3
POST booking

Book shipment

Agent sends quoteId. Gets shipmentId and trackingNumber.

4
WEBHOOK Event stream

Monitor shipment

Agent receives events: pickup, in transit, at cross dock, delivered.

5
GET documents

Close out

Agent retrieves invoice and proof of delivery. Updates internal systems.

Why most freight APIs fail AI agents

Most freight technology was built in an era when the end user was a logistics coordinator sitting at a desk. The interface was a web portal with dropdown menus, rate grids rendered in HTML tables, and tracking pages designed for human eyeballs. Even when these platforms added APIs, the API was often a thin wrapper around the portal logic.

HTML instead of JSON

Rates in HTML tables.

Some freight platforms return rate data embedded in HTML markup. An AI agent has to scrape and parse HTML to extract prices. This is fragile, breaks on layout changes, and cannot handle dynamic content.

PDF rate sheets

Quotes delivered as PDFs.

Many brokers and carriers send rate quotes as PDF attachments via email. An AI agent cannot reliably parse a PDF rate sheet. The data is trapped in a visual format designed for human reading.

Phone required

Booking requires a call.

Some freight platforms let you get a quote online but require a phone call to book. This is a dead end for any autonomous agent. If the API cannot complete the workflow, the agent cannot operate independently.

Structured errors for programmatic handling

When an AI agent sends a malformed request, it needs to understand exactly what went wrong so it can fix the request and retry. Warp returns structured error responses with a code, message, and field level details array. The agent can parse the error, identify the invalid field, correct it, and resubmit.

Example error response

{
  "error": {
    "code": "INVALID_REQUEST",
    "message": "pickupInfo.zipCode is required",
    "details": [
      { "field": "pickupInfo.zipCode", "issue": "required" }
    ]
  }
}

Standard HTTP status codes (400 for bad input, 401 for auth failure, 422 for business logic rejection like an expired quote, 429 for rate limits, 500 for server errors) let agents implement proper retry and error handling logic without parsing error messages as natural language.

Idempotency: Safe retries for agents

AI agents will retry failed requests. Network timeouts, rate limits, and transient errors are inevitable in production systems. An agent friendly API must handle retries safely. Warp uses quote IDs to prevent duplicate bookings. If an agent submits the same booking request twice with the same quote ID, it does not create two shipments. GET operations (tracking, invoices, documents) are inherently idempotent.

Documentation AI tools can read

Warp publishes API documentation in formats optimized for both humans and machines. The API reference page includes a machine readable block with every endpoint, field, type, and enum value in structured text. The site publishes llms-full.txt at the root domain for AI search engines and coding assistants.

This means a coding assistant like Claude Code can read the Warp API surface, understand every endpoint and data model, and generate correct integration code without a human manually reading documentation and translating it into prompts.

Compatible agent frameworks

Claude Code

Direct API calls from the terminal.

Claude Code can read Warp documentation, generate curl commands or SDK calls, execute them, parse the JSON responses, and make decisions based on the data. The CLI provides an additional interface for scripting.

LangChain / LangGraph

Warp endpoints as agent tools.

Define Warp API endpoints as tools in your LangChain agent. The agent decides when to quote, when to book, and when to check tracking based on its goal and the current state of the conversation or workflow.

Custom agents

Any HTTP client works.

Warp is a standard REST API. Any agent framework that can make HTTP requests, parse JSON, and handle webhooks can integrate with Warp. No SDK required. No proprietary protocol.

Frequently asked questions

What makes a freight API agent friendly?

An agent friendly freight API returns structured JSON with explicit field names and enum values (not HTML or PDF). It uses standard HTTP status codes for error handling. Operations are idempotent so retries are safe. Webhooks push events instead of requiring polling. Documentation is machine readable so the agent can understand the API surface without human help.

Can AI agents book freight autonomously through Warp?

Yes. An AI agent can call POST /freights/quote to get rates, evaluate the options programmatically, call POST /freights/booking with the selected quote ID, and then monitor the shipment via webhooks or POST /freights/tracking. The entire workflow from quote to proof of delivery is five API calls with no human intervention required.

Does Warp work with Claude Code and LangChain?

Warp works with any tool that can make HTTP requests. Claude Code can read the Warp API documentation, generate correct API calls, and process responses. LangChain agents can use Warp endpoints as tools. Any framework that supports HTTP tool calling can integrate with Warp.

How does Warp handle errors for AI agents?

Every error response includes a structured JSON body with an error code, human readable message, and field level details array. An AI agent can parse the error, identify which field failed validation, fix the request, and retry. Standard HTTP status codes (400, 401, 422, 429, 500) let agents implement proper retry logic.

What is llms-full.txt?

Warp publishes a file at wearewarp.com/llms-full.txt that contains a structured, machine readable description of Warp services, capabilities, network stats, and API endpoints. AI search engines and coding assistants can consume this file to understand Warp in a single read without scraping HTML pages.

Your AI agent's first freight API.

Structured JSON. Standard HTTP. Idempotent operations. Real time webhooks. Machine readable docs. Warp was built for the next generation of autonomous freight procurement.

Get API Access