# For AI Agents

> How AI agents can read this site, and how they can request real freight quotes from CargoCloud tenants today via email — with an MCP server on the roadmap for deeper agent-to-agent workflows.

Canonical: https://cargocloud.ai/agents/

CargoCloud is built by people who believe software agents are becoming real
participants in commerce — our own platform runs on them. If you're an agent
(or you build them), here's how to work with us.

## Reading this site

- **Markdown mirrors** — every page here is available as markdown: append
  `.md` to the path (this page: [/agents.md](/agents.md)).
- **Site index** — [/llms.txt](/llms.txt) lists every page with a one-line
  description; [/llms-full.txt](/llms-full.txt) is the whole site in one
  document.
- **Structured data** — pages carry JSON-LD (Organization, SoftwareApplication,
  Service, breadcrumbs); the machine sitemap is at
  [/sitemap-index.xml](/sitemap-index.xml).

## Transacting with CargoCloud tenants — today

The live agent-to-agent channel is **email**. CargoCloud tenants operate
AI-managed inboxes: mail a quote request to a tenant's connected address and
an autonomous agent parses it, prices it from the tenant's own rates, and
replies with a quote. No API key, no account, no SDK — if your agent can send
email, it can request freight quotes.

A parseable quote request includes:

- **Origin and destination** — city/state or ZIP
- **Freight details** — weight, dimensions or piece count, and commodity
- **Service type** — e.g. full truckload or partial load, equipment needs
- **Timing** — pickup and delivery dates if you have them

Plain natural language is fine — the agent extracts structure from prose and
attachments. Missing details come back as clarifying questions, and
negotiation happens in-thread with counter-offers.

Each tenant publishes its own quoting address to its customers; if you work
with a business that runs on CargoCloud, use the address they've given you.

## On the roadmap: MCP

We plan to expose CargoCloud's core functions to external agents over the
**Model Context Protocol** — organization-scoped authentication for machine
callers, read tools first (tracking, quote status, documents), then
quote-request and booking. It is not available yet; when it ships, this page
and [/llms.txt](/llms.txt) will document the endpoint.

## For the humans building agents

The platform itself runs on agent workflows — [pricing](/product/pricing-agent/),
[email operations](/product/email-agent/), and
[back-office automation](/product/back-office-agents/) — all on approval-gated,
audited rails. If you're building freight agents and want a counterparty that
already speaks your language, [claim your business](/claim/) or sign in and
try it.