Your Ockam agents have built-in memory that allows them to maintain state across prompts: 
> What is your name?

(in a pirate accent) Ah, me name be Captain Jack Sparrow,
the infamous pirate captain of the Black Pearl. Savvy?

> Where do you live?

(laughs) Ah, me home be the sea, matey. Me ship, the Black Pearl,
be me abode. But if ye be lookin' fer a place to find me on land,
I've been known to frequent the island of Tortuga. It be a lawless
place, full of scurvy dogs and scallywags, but it be home to me.

> What have I asked you before?

(chuckles) Ah, ye be askin' about yer previous questions, eh?
Alright then, matey... Ye asked me what me name be, and then ye
asked me where I live. That be the extent o' our conversation
so far, savvy?

>

​
Isolating conversations

You can have multiple separate conversations with an agent at the same time by using a different conversation identifier for each one: 
await agent.send("What have I asked you before?", conversation="1")

​
Isolating conversations per user

You can add an extra layer of isolation by using a unique scope identifier (like a customer ID) when conversation-level isolation alone doesn’t meet your needs in multi-user environments: 
await agent.send("What is 56 divided by 27?", scope="user-1", conversation="1")