Skip to main content
The Nash agent doesn’t just remember individual facts. Over time, it also forms a sense of how you work — the kinds of deliveries you watch, the questions you keep coming back to, the way you like to follow up. The next time you start a conversation, the assistant arrives already a little oriented to you, instead of starting from a blank slate every time. You never write this picture, and you don’t have to maintain it. Nash builds and refreshes it on its own, in the background, so the agent gradually feels more tuned to your work without anyone curating notes. This page is a closer look at one piece of Memory and learning — the “your profile / working style” part — viewed from the angle of “how does Nash get to know me?” For who controls memory and learning, what else it remembers, and the privacy considerations, start with that page.

​
What it produces

The result is a short, plain-language picture of how someone tends to work — the topics that come up again and again, the parts of an operation a person focuses on, the way they like to follow up. Nash uses this picture quietly: it folds into the background context the agent has when you begin a conversation, so the assistant is better oriented to you from the first message. This picture is inferred, not authored. Nash reads your recent activity and summarizes the patterns it sees. That’s different from Knowledge, which is reference material you type in by hand, and it’s part of the “learning” side of Memory and learning — the context the agent builds for itself.
This is behind-the-scenes personalization. The inferred picture isn’t a document you open and read line by line — it’s working context for the agent. Its only visible payoff is that the assistant feels like it already knows how you work.

​
How it gets built

Nash forms this picture by periodically reflecting on your recent conversations rather than your entire history, so it stays current with how you’re working now — not how you worked months ago. It refreshes on its own, in the background; there’s nothing for you to schedule or trigger. Because the work runs in the background, it never interrupts a conversation and you don’t wait on it. Whatever the most recent reflection produced is what the agent draws on the next time you chat. The picture Nash forms about you is yours — it reflects your conversations, not a blended view of your whole organization. Two people on the same team can have very different working-style pictures, because each is built from that person’s own activity.

​
What you can shape

Most of this is intentionally quiet. The picture Nash infers is working context for the agent, not a page you browse. But there is one part you can author directly: a short profile about yourself — your role, what you’re responsible for, and the kinds of things you care about most. That profile is yours to view and edit from your profile settings. It’s the part of the picture you control by hand, and it gives the agent a reliable anchor alongside the patterns it picks up on its own.
LayerWho creates itCan you edit it?
Your authored profileYou — your role, responsibilities, and focus areasYes, from your profile settings
The inferred working-style pictureNash, from your recent conversationsNo — it’s built and refreshed automatically
Keeping your authored profile current is the most direct way to steer how Nash understands you. A clear “here’s my role and what I care about” gives the agent a reliable anchor, and the patterns it infers fill in around it.

​
How this fits with memory and learning

Forming a working-style picture is one of the ways the agent gets smarter about you over time. It sits alongside the other things Memory and learning covers — remembering useful facts about you, keeping track of a conversation’s goals, and noting context about the orders and drivers you work with. A couple of things are worth keeping straight:
  • The agent does the work. You don’t author the inferred picture, and there’s nothing to curate — Nash forms and refreshes it itself. The only part you author by hand is your own profile.
  • It’s about you, not the whole org. The inferred picture reflects your conversations. The broader memory-and-learning controls — what gets remembered and the master controls over it — are managed for your organization by an admin, as described on the Memory and learning page.
For the full picture of what gets remembered, who controls it, and the privacy considerations, see Memory and learning.

​
FAQ

The inferred working-style picture is background context the agent uses, not a document surfaced for you to read. The part you can view and edit directly is your authored profile — your role, responsibilities, and focus areas — in your profile settings.
No. Knowledge is reference material you write and control by hand — SOPs, policies, links. A working-style picture is inferred by Nash from how you actually work. Knowledge is fixed until you edit it; the inferred picture evolves on its own.
No. The inferred picture reflects your conversations and is about you. The broader controls over what the agent remembers are managed at the organization level by an admin — see Memory and learning.
The agent forms the inferred picture on its own, and you control your authored profile by hand. The organization-wide controls over what gets remembered are an admin setting — see Memory and learning for who can change them.

Memory and learning

What lets the agent remember and learn across conversations — and the controls that govern it.

Knowledge

The reference material you author by hand — distinct from anything the agent infers on its own.