# Overview

Carve is our new territory planning AI chat agent. With Carve, you'll be able to model multiple different territory scenarios in just a few minutes.

### How to Access Carve

Carve is accessible in the Gradient Works web app at <https://app.gradient.works/carve>. You must have an active Gradient Works account to access this. (Contact us if you can't get in.)

Once you're logged in, Carve is located under the square icon with a slash through it.

<figure><img src="/files/W3okWLkbVxJOGeasUpI8" alt="" width="375"><figcaption></figcaption></figure>

### Basic Concepts

There are a few major elements to Carve.

* **Project**: A project is your space for testing different carving scenarios for a set of accounts. A project might be something like "Mid-Market Territories 2026".
* **Scenario**: Each scenario is one option for how accounts may be distributed, so you can compare different approaches to find the best one for your team. One project will likely have several different scenarios where you model the impact of different requirements. [More on scenarios here](/kb/carve/scenarios.md).
* **Account sheet**: The list of accounts you're trying to carve, with important account data to help inform your carves. This can be created from a Salesforce or HubSpot report, or from a CSV. [More on account data ](/kb/carve/account-data.md)here.

### Getting Started

To get started, you'll first create a project. Once you have a project created, you can add an account list and run through various territory scenarios.

1. **Create a new project** using the big *+New Project* button in the top right corner. Give it a name.
2. **Import account data**. You can import Excel and CSV files, or import from a Salesforce or HubSpot report. This should be a list of the accounts you want to design territories for, including columns for relevant account data, like employee count, revenue, account type, etc... You can [update your account sheet](/kb/carve/account-data.md#updating-account-sheet-data) with additional data throughout the process.
3. **Create a new scenario**. Click the *+New* button on the left side of the chat agent to create a new scenario. A scenario is one set of requirements you want to model. You'll be able to tweak and test more scenarios later.
4. **Tell the agent about your requirements.** What kind of territories are you trying to carve with this scenario? What data should it prioritize? Tell it as little or as much as you'd like to start the conversation, and it'll ask follow up questions to clarify any relevant details. There's more [here on creating effective scenarios](/kb/carve/scenarios.md#creating-effective-scenarios).
5. **Refine the scenario's details**. The agent will ask clarifying questions. You can choose to answer them yourself or have the agent decide.
6. **Run the scenario**. When you're happy with the details, tell the agent to run the scenario. This will take a few minutes, and then it will come back a description of what it did, a table that summarizes the results, and an updated account sheet with new columns that detail the carve.
7. **Tweak your scenario**. You can make small updates to this scenario, import updated account data, or add overrides to your results before finalizing them.
8. **Deploy your territories or run a new scenario**. After you get results, you can deploy your territories to CRM, or export your results for deeper analysis. Or you can create a new scenario to try another or refined set of requirements. [More on deploys here](/kb/carve/scenarios.md#deploying-your-results).

<figure><img src="/files/OZx1QNPaqlq7wNUDI4HL" alt=""><figcaption></figcaption></figure>


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.gradient.works/kb/carve/overview.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.
