> ## Documentation Index
> Fetch the complete documentation index at: https://docs.firedrill.run/llms.txt
> Use this file to discover all available pages before exploring further.

# Enterprise

> A synthetic version of your business, scoped with the Firedrill team, for evaluating AI agents before they get access to live systems.

Firedrill Enterprise gives your organization a reusable synthetic version of your business. Your team can run simulations there, compare agents on the same tasks and evaluate how they behave, without giving them access to live systems or real customer data.

It is set up with the Firedrill team and is not available through self-serve sign-up. To discuss it, visit [firedrill.run/enterprise](https://firedrill.run/enterprise).

## What the environment represents

| Part | What it covers |
| - | - |
| **Tools and systems** | Simulated apps and services your agents work in. |
| **Data** | Synthetic records and the relationships between them. |
| **Users and roles** | People with goals and context. |
| **Permissions** | Access levels and the actions each role may take. |
| **Workflows** | Tasks, decisions and handoffs. |
| **Policies** | Your business rules and requirements. |

## How it works

**1. Define your world**

Scope the systems, data, roles and policies with the Firedrill team.

**2. Set tasks and requirements**

Agree what agents should do and what must always hold.

**3. Run agents**

Run vendor and internally built agents in the same world, on the same tasks.

**4. Inspect outcomes**

Review actions, resulting state and evidence against each requirement.

## When teams use it

* **Before buying.** Bring shortlisted vendors into the same business context and see where each meets your requirements.
* **Before granting access.** Explore exceptions, permissions and handoffs with simulated services and synthetic records before connecting an agent to production.
* **After a change.** Rerun saved scenarios when the model, prompt, tools or policies change, and compare with your baseline.

Typical questions include which agent resolves payment disputes within your approval rules, which catches missing information before opening an account and which respects access limits when updating customer records.

## Start with one workflow

Most teams start with one workflow. Together with the Firedrill team, you agree the systems to represent, the scenarios to run and the evidence you need to make a decision. Coverage and setup depend on the systems and workflows involved.

> [card] [Talk to the team](https://firedrill.run/enterprise)
>
> Discuss your enterprise environment.

> [card] [Already building an agent?](https://docs.firedrill.run/quickstart.md)
>
> Start with the CLI and synthetic tools today.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.