How to Connect Box to an AI Agent: Step-by-Step Guide
· DoubleX
Quick answer
To connect Box to an AI agent, add the Box connector, authorize your account, and choose which actions the agent may use. In DoubleX Agentes, you connect Box under Connectors, assign it to an agent with read-only permissions, then ask your first question and review the task history.
Box holds the contracts, reports, and shared folders your team relies on, but finding the right file still takes manual searching. Connecting Box to an AI agent lets you ask for files in plain language. This guide walks through the full path, from account creation to your first request, with a focus on permissions and access control.
What can an AI agent do with Box?#
An AI agent connected to Box can search and read files and folders, using Box AI, and answer questions based on what it finds. The Box connector in DoubleX Agentes is built around searching and reading, which makes it a good fit for a cautious first setup.
An AI agent is an assistant that uses a language model plus tools, here the Box connector, to do a task for you instead of only chatting. For ideas on what to ask, see What You Can Automate in Box with AI: Example Requests. This guide covers only the connection itself.
What do you need before connecting Box?#
You need a DoubleX Agentes account, access to the Box account you want to use, and a clear idea of which folders the agent should reach. A few minutes of preparation avoids permission surprises later.
| Item | Why it matters |
|---|---|
| DoubleX Agentes account | Creating one is free and requires no credit card |
| Box login | You authorize the connection with your own Box account |
| Target folders | The agent can only see what the authorized Box account can see |
| An agent | The assistant that will use Box, which you can create or generate from the diagnostic |
| Decision on scope | Personal, team, or company connection (see the last section) |
How do you connect Box in Connectors?#
You connect Box from the Connectors screen by adding Box, authorizing your account, and then assigning the connector to the agent that will use it. Here is the full flow:
- Create your account and, if you like, run the diagnostic so you have an agent ready.
- Open Connectors and choose Box from the list of available systems.
- Authorize access with your Box account when prompted.
- Choose the scope of the connection: personal, team, or company.
- Open Agents, select the agent that should use Box, and add the Box connector to it.
- Set which Box actions the agent may use (next section).
- Send a first request in a task and check the result.
Credentials for connections are stored encrypted with AES-256-GCM, so tokens are not kept in plain text. If you want to manage connectors programmatically, see the Connectors API documentation. New to the platform? The Quickstart covers the basics.
How do you limit which actions the agent can use (read-only)?#
For each agent, you choose which systems it can use and, within each system, which actions it can run. Actions that change data can stay turned off, so the agent only reads.
For Box, the connector is focused on searching and reading files and folders, so a read-only setup is a natural default. Keep these principles in mind:
- Give each agent only the systems it needs. A finance agent may need Box; a social media agent may not.
- Leave actions that change data switched off unless you have a specific reason to enable them.
- Remember that before sending, deleting, or changing anything in your systems, the agent asks for confirmation and shows the step-by-step of what it did.
How do you make the first request and check the history?#
Open a task with the agent and ask something simple that requires finding a file. Every request becomes a task in the Tasks screen, with the full history: your request, each tool the agent used, and the result.
Copyable first requests:
- "Search Box for the most recent version of the Q3 vendor contract and tell me the file name and folder."
- "List the files in the Marketing Assets folder in Box."
- "Find the onboarding checklist in Box and summarize it in five bullet points."
- "Which files in the Legal folder mention the word renewal?"
After the agent replies, open the task and review which Box tools it called and what it returned. The history shows who asked, what was done, and when, which is useful for auditing and for tuning your agent's instructions. You can also add company context to the Brain so the agent knows how your folders are organized.
Should you use a personal, team, or company connection?#
Choose personal if the access should belong to one person, team if a group shares it, and company if every authorized agent should use the same connection. A connection can be personal (valid only for that person, such as their own email), team, or company.
| Scope | Best for | Consideration |
|---|---|---|
| Personal | Individual files and your own Box account | Only works for you |
| Team | A department sharing a set of folders | Limit the account to the team's folders |
| Company | Shared company-wide documents | Use an account with deliberately narrow access |
Roles of owner, admin, and member, along with sharing by person, team, or company, help you control who manages these connections. Routines that run automatically use the permissions of the person who created them, so create them with the right account.
Frequently asked questions#
Can the agent edit or delete my Box files?#
The Box connector is designed for searching and reading files and folders. You also control which actions each agent can use, and the agent asks for confirmation before sending, deleting, or changing anything.
Do I need to know how to code to connect Box?#
No. You connect Box from the Connectors screen and describe what you need in your own words. Developers who want more control can use the REST API, the CLI, or the MCP server.
Is my Box content used to train AI models?#
No. Your company's content is not used to train models, and requests only go to AI providers that don't use what they receive for training. You can also enable zero retention in the Models screen.
Where can I see what the agent did with Box?#
In the Tasks screen. Each request is a task with its full history: the request, each tool the agent used, and the result.
Can the agent check Box on a schedule?#
Yes. You can set up a routine that runs at a set time, on a webhook, or when an email arrives. See Triggers for how to configure them.
Related guides
- How to Connect Jam to an AI Agent: Step-by-Step GuideLearn how to connect Jam bug reports to an AI agent in DoubleX Agentes: set up the connector, pick who can use it, limit permissions, and run a first test.
- How to Connect Hex to an AI Agent: Step-by-Step GuideLearn how to connect Hex to an AI agent: set up the connector, choose personal, team or company scope, limit permissions and run your first data question.