> ## Documentation Index > Fetch the complete documentation index at: https://felan.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # How it works > How Felan turns requests and events into contextual, delegated SDLC work. Felan is a stateful AI SDLC agent. A team can invoke it from the web dashboard, a connected conversation, a schedule, or an external-system event. Felan combines the request with the context available to that team, then completes focused work directly or delegates specialized work. ## The interaction loop ```text theme={null} Request, schedule, or event → Felan reads the available team and repository context → Felan works directly or delegates specialized work → Progress and results appear in a session → A teammate follows up or automates the next request ``` ## Sessions keep the work together A [session](/docs/platform/sessions) contains the prompt, live transcript, status, and delegated agent activity for one body of work. Conversational sessions support follow-up messages, so the team can refine the outcome without starting over. Requests from connected conversation threads resume the corresponding Felan conversation. Automation runs create sessions with source information so the team can trace why the work started. ## Context is team-scoped Felan uses only context available to the current team and request. Depending on configuration, that can include: * The current message and earlier messages in its conversation * Durable team context and learned memory from [Knowledge](/docs/platform/knowledge) * Uploaded team reference files * Repositories authorized through source-control integrations * Event payloads and metadata from connected systems * Environment names and variables configured for the team Felan discovers repository structure and conventions before acting instead of assuming a framework or workflow. ## One main agent, specialized help The main Felan agent communicates with the team and coordinates the result. It can inspect repositories, make focused changes, run commands, and manage platform resources exposed by its tools. For broader, specialized, or parallel work, Felan can delegate to agents for areas such as architecture, implementation, review, QA, investigation, and memory consolidation. Delegated agents run asynchronously and share the root session workspace, so the main agent can continue coordinating while they work. Each completion returns through the parent session. ## Automations reuse plain-language requests An [automation](/docs/platform/automations) combines: 1. A name 2. One or more schedule or event triggers 3. Plain-language instructions describing what Felan should do When any enabled trigger fires, Felan starts a session with those instructions and the event context available to the team. Automations are useful when the desired response is repeatable, such as reviewing a recurring signal, checking a deployment event, or preparing a scheduled report. ## Access follows configuration Felan’s available evidence and actions depend on connected repositories, configured integrations, selected agent skills, team environments, and the current user’s request. If required context or access is unavailable, Felan should identify the blocker rather than inventing a result. Start work, inspect progress, and continue a conversation. See how teams connect planning, implementation, verification, and delivery.