How we built Fern Agent

Fern Agent is a documentation-focused AI agent that investigates analytics, search activity, configuration, and user feedback to identify underperforming pages and propose or make edits through a unified conversation available in Fern’s Dashboard, Slack, and MCP-compatible external coding agents. Although each interface is tailored to its native workflow, they share the same tools, persistent organization-level conversation state, and durable backend workflows, allowing tasks to survive crashes or deployments without repeating completed actions. Its editing system uses a fast GitHub API-based lane for simple changes and escalates to a persistent full-checkout session for more complex work, delivering updates either to the visual editor as drafts or to GitHub branches and pull requests. Fern Agent executes under the requesting user’s authentication rather than a privileged service account, enforcing existing role-based access controls and screening all external or user-provided content for prompt-injection attempts before it reaches the model. Built collaboratively by Fern’s team and now available to customers, the product emphasizes evidence-based investigation, human review of changes, shared tools for both hosted and customer-operated agents, and Git-based controls over published documentation.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.