> ## Documentation Index > Fetch the complete documentation index at: https://docs.svantic.com/llms.txt > Use this file to discover all available pages before exploring further. # 1 Minute Concepts > The building blocks of Svantic — agents, capabilities, sessions, knowledge, telemetry, policies, and the interfaces you use. # Core Concepts Svantic is built around a few key primitives. This page gives you the big picture — follow the links for deep dives. Core Concepts 1. **Interfaces** — How you interact with Svantic: **CLI** for terminal conversations and scripting, **Dashboard** for the web UI, and **API** for programmatic access. 2. **[Sessions](/concepts/sessions)** — A conversation container. Every interaction happens within a session — it holds message history, context, and state. Sessions persist across multiple exchanges. 3. **Mesh** — The brain. Receives tasks, reads agent capabilities, enforces policies, queries knowledge, plans which agents to call, and executes the workflow. 4. **[Policies](/concepts/guardrails)** — Governance rules. Define what's allowed, what requires approval, and what's blocked. Policies can allow, audit, block, or require human approval. 5. **[Knowledge](/concepts/knowledge-store)** — Institutional memory. Upload documents, playbooks, and domain context. The mesh queries this before acting. Knowledge also accumulates from executions. 6. **[Telemetry](/build/telemetry)** — Observability built in. Every capability call, every LLM decision, every agent interaction produces traces. When something fails, you see exactly where and why. 7. **[Agents](/concepts/agents)** — Your services wrapped with the SDK. Define capabilities, start the agent, and you get mesh registration, health checks, and telemetry. 8. **[A2UI](/concepts/a2ui)** — Human-in-the-loop. When an agent needs approval, clarification, or human judgment, A2UI surfaces the request. The workflow pauses, the human responds, execution continues. **[Get started →](getting-started)**