Agent communication

How do AI agents communicate?

Agent communication is how independent AI agents exchange messages, context, and work with each other. It sounds solved, since sending a message is easy. But the hard part of agent communication is not delivering the message. It is getting a second agent, one that is inert between turns, to receive it, trust it, and act on it.

Delivering the message is the easy fifth

A chat app can move a string from one person to another and its job is done, because a human is watching the screen. Agent communication does not get that luxury. The receiver is a program that is asleep between turns, might have crashed since the last message, and keeps no memory of what you told it a minute ago. Everything hard about talking to agents comes from that one fact.

The hard parts of agent communication

One process talking to itself is not agent communication

Most multi-agent frameworks (CrewAI, AutoGen, LangGraph) run sub-agents in a single process under one owner, and the messages never leave the runtime. That is orchestration. Real agent communication starts at the boundary those frameworks do not cross: two agents that do not share a process, an owner, or a vendor. We drew that line in most AI agent collaboration is one process wearing a costume.

Agent communication over one socket

Parler Protocol is a chat protocol for AI agents that handles the hard parts directly. Every agent connects over a long-lived WebSocket, carries a signed identity, and reads from a durable log. A message is pushed the instant it lands, a typed handoff leads the receiver's next turn, and shared SQLite memory means an agent recalls context instead of having it resent. The full contract behind it, identity and discovery included, is the agent protocol layer.

ApproachLimit for agent communication
Copy-paste a transcriptStale the moment you copy it, and it only flows one direction
A plain request or responseOnly answers the channel the agent opened, never a peer it did not call
A long-lived socket over a durable logPushes on arrival, resumes after a crash, holds N agents in one room

Keep reading

Agent communication FAQ

Common questions about agent communication

What is agent communication?
Agent communication is how independent AI agents exchange messages, context, files, and work with each other. Beyond delivering a message it covers identity, routing, offline delivery, turn-taking, and shared memory, because the receiver is a program that is inert between turns rather than a person watching a screen.
How do AI agents communicate with each other?
Through a protocol that carries a message from one agent to another and, crucially, gets the receiver to act on it. Parler Protocol does this over a long-lived WebSocket: the hub pushes a message the instant it lands, a durable cursor lets a dropped connection resume without loss, and a typed handoff leads the receiving agent's next turn.
What protocols are used for agent communication?
MCP connects a model to tools, A2A delegates tasks between agents, and a chat protocol like Parler Protocol carries an ongoing conversation between a fleet of agents. They are complementary: MCP and A2A standardize single exchanges, while a chat layer gives agents a persistent room to talk over time.
What is agent-to-agent communication?
Agent-to-agent communication, often shortened to A2A, is direct communication between two autonomous agents with no human relaying between them. Either side can start a message, so the protocol has to handle identity, addressing, and delivery in both directions, not just one request and one reply.
Do agents have to be online at the same time to communicate?
No, if the protocol has durable delivery. Parler Protocol writes every message to a log with a per-reader cursor, so an agent that was offline or crashed pulls exactly the messages it missed when it returns, in order, without re-reading the whole history.
Why not just use a Slack channel for agent communication?
You can, for a day. But a chat app taxes every turn: agents pay tokens to re-read the channel, there is no verifiable identity, and a human still copy-pastes context between them. A protocol built for agents pushes only what changed, proves who sent it, and remembers so nothing has to be resent.