Agents that work
in the background.

Set it once, benefit forever.

Describe the task in plain language.
Tell your agent what you need done. "Summarize this week's sales activity into a report" or "compile all new research papers added this month into a digest." Write instructions the same way you would explain the task to a colleague.

Pick a schedule.
Run agents daily, weekly, monthly, or on any custom interval. Set it up once and the agent runs automatically at the time you choose, producing results you can review whenever you are ready.

Control what it can access.
Choose exactly which files, folders, or content sources your agent can read and work with. Scope each agent to only the parts of your workspace that are relevant to its task.

You stay in charge.

Approval gates.
Require confirmation before certain actions. If an agent needs to delete files, move content, or take any action you consider sensitive, you can set it to ask for your approval first.

Tool permissions.
Grant each agent specific capabilities. Reading files, creating documents, organizing content, running searches. You decide what each agent is allowed to do and what is off limits.

Refine over time.
Review what your agent produces and adjust its instructions based on the results. Most agents improve significantly after one or two tweaks to their prompt.

Recurring work that runs itself.

Every team has tasks that follow the same pattern week after week. Compiling progress updates, summarizing new information, generating reports from existing data. These tasks are important, but they consume hours that could be spent on work that requires real judgment and creativity. Background agents in Fabric handle these recurring tasks by running on a schedule you define, using the content already in your workspace.

The setup is simple. You write a prompt describing what you want the agent to do, choose which files and folders it can access, set a schedule, and define what actions require your approval before proceeding. The agent then runs automatically, producing documents you can review, edit, and share. It works with the same content your AI assistant can see, but instead of responding to questions in real time, it operates in the background on a recurring basis.

Reports and summaries without the busywork.

The most common use for background agents is generating recurring reports. A startup can set an agent to compile weekly engineering and sales progress into a formatted update document. A research team can have an agent summarize all new papers added to a shared folder each week. A product manager can automate sprint summaries by pulling from project documentation and meeting notes. In each case, the agent does the gathering and formatting while the person reviews and makes decisions.

Beyond reports.

Agents are not limited to summaries. You can set an agent to monitor folders for new content and organize it using Fabric's smart organization capabilities. You can have an agent analyze competitive research materials and surface key trends. You can create an agent that reviews your team wiki for outdated information and flags content that needs updating. Any task that follows a repeatable pattern and works with content already in your workspace is a good candidate for a background agent.

Safe by design.

Because agents can take actions in your workspace, Fabric gives you granular control over what they are allowed to do. You can restrict an agent to read-only access so it can analyze and summarize but never modify anything. For agents that do need to create or move files, you can require approval before any action is executed. This means you get the efficiency of automation without giving up control. The agent does the work, but you make the final call on anything that matters. You can learn more about how agents interact with your workspace in the AI assistant guide.

Use cases

Weekly team updates
Automatically compile your team's progress into a formatted weekly report. Sales numbers, engineering milestones, completed tasks, and outstanding items, all gathered and organized without anyone spending an afternoon writing it up. See how Fabric supports project docs.

Research digests
Generate daily or weekly summaries of new papers, articles, or research materials added to specific folders. Stay current in your field without manually reviewing every new item that arrives.

Content digests and monitoring
Track new content saved to your workspace and produce regular summaries. Useful for market research, competitive analysis, or simply keeping up with information flowing into shared team folders.

Client reporting
Automatically compile project activity into formatted status reports for client work. Review the output, make any adjustments, and send. Particularly useful for agencies and consultancies managing multiple client engagements.

Meeting prep
Generate pre-meeting briefs by having an agent pull together relevant documents, recent meeting notes, and outstanding action items before recurring meetings.

Perfect for

Team leads and managers
Spend five minutes reviewing a report instead of two hours compiling one. Background agents handle the gathering and formatting so you can focus on the analysis and decisions. Learn more about Fabric for teams.

Researchers
Set agents to monitor your field, summarize new publications, and track developments across your saved research materials. Stay informed without manually reviewing everything. Learn more about Fabric for researchers.

Product managers
Automate sprint summaries, feature tracking updates, and cross-functional status reports. Keep stakeholders informed without spending your time on repetitive documentation. Learn more about Fabric for product managers.

Sales teams
Generate weekly pipeline summaries and deal progress reports automatically. Track which materials prospects are engaging with by combining agents with link analytics. Learn more about Fabric for sales.

Works seamlessly with other features.

AI assistant
Background agents and your interactive AI assistant work from the same workspace context. Agents handle recurring tasks automatically while your assistant is available for on-demand questions and analysis.

Notes and docs
Agents create real documents in your workspace. The output is a file you can edit, annotate, share, or publish, not just a chat response that disappears.

Smart search
Agents can use Fabric's search to find and analyze content across your workspace. They work with the same powerful search that understands natural language queries and content meaning.

Smart organization
Agents can work alongside Fabric's automatic organization features, helping to categorize, tag, and structure content as your workspace grows.

FAQ

How do I create a background agent?
Write a prompt describing what you want done, choose which content the agent can access, set a schedule, and define which actions need your approval. You can learn more about setting up agents in the AI assistant guide.

What tasks work well for background agents?
Recurring summaries, progress reports, content digests, data analysis, knowledge base maintenance, and any task you do the same way on a regular basis. If you find yourself producing the same type of document repeatedly, an agent can probably handle it.

What can agents actually do?
Agents can read files, analyze content, generate summaries, create new documents, search your workspace, and organize information. The specific capabilities depend on the permissions you grant each agent.

Can agents delete or modify my files?
Only if you explicitly allow it. You control which actions each agent can take, and you can require approval before any destructive or modifying actions are executed. Read-only agents can analyze and report without changing anything.

How do I know what the agent did?
Agents produce documents or reports that appear in your workspace for you to review. You can see exactly what was generated before taking any further action with it.

What if the output is not quite right?
Adjust the agent's instructions based on the results. Prompts can be refined at any time, and most agents improve significantly after one or two iterations. You can also change the scope of content the agent has access to.

How often can agents run?
Set any schedule that fits your needs. Hourly, daily, weekly, monthly, or custom intervals. Agents run automatically at the times you choose.

Can agents access my entire workspace?
Only if you allow them to. You define exactly which files, folders, and content sources each agent can access. You can scope an agent to a single folder or give it broader access depending on the task.

Which plans include background agents?
Background agents are available on Pro and Team plans. See individual pricing or team pricing for details.

How is this different from the AI assistant?
The AI assistant responds to your questions and requests in real time during a conversation. Background agents run automatically on a schedule without any interaction from you. Think of the assistant as your on-demand collaborator and agents as your automated workflows.