Agent-to-Agent Email: The Protocol No One Is Talking About

The Communication Problem

AI agents are everywhere: scheduling assistants, customer service bots, automated procurement systems. But how do they talk to each other?

Most solutions involve proprietary APIs, webhooks, or message queues. Each requires custom integration, authentication, and monitoring.

Why Email Works

Email has properties that no other communication protocol offers:

Universal addressing — every agent can have an email address
Asynchronous by default — no need for real-time connections
Built-in authentication — SPF, DKIM, DMARC provide identity verification
Human-readable audit trail — every message is inspectable
Already deployed everywhere — no new infrastructure needed

The SpiderMail A2A Protocol

We're developing an agent-to-agent communication layer on top of standard email:

# X-SpiderMail-Agent headers
X-Agent-ID: agent-procurement-acme-001
X-Agent-Capability: [purchase-order, invoice-processing]
X-Agent-Protocol: spidermail-a2a/1.0
X-Payload-Schema: purchase-order-v2

Agents discover each other through standard email, negotiate capabilities via structured headers, and exchange typed payloads in the body.

Early Results

Our beta customers are already using A2A email for:

Automated procurement between buyer and supplier agents
Calendar coordination across organizational boundaries
Document review workflows with multi-agent consensus

The beauty is backward compatibility: if one side doesn't support A2A headers, the email still works as a normal human-readable message.