# What Dify Is — Dify Workflow Basics

Source: https://www.skillbyai.com/en/dify-workflows/i-what

> A visual builder and runtime for LLM applications.

## Build, run and operate in one place

**Dify** is an open-source platform for building applications on large language models. It combines a visual **workflow** editor, **prompt** management, **knowledge bases** for retrieval-augmented generation, **tools** and plugins, **agents**, logs and analytics, and an **API** plus a ready-made web app for each published application. It connects to many model providers (hosted APIs and self-hosted models) so you can switch models without rewriting the app. You can use the hosted cloud service or **self-host** it with Docker. Dify's node names, menus and options change between versions; check the current Dify documentation.

## An open-source platform for LLM apps

Dify lets you build, test and publish LLM applications visually, connected to many model providers.

![Three ideas: what Dify is, app types, setup.](assets/figures/dify-workflows/section-1-map.svg) — Figure 1.1 — Platform, app types and setup.

## A workshop with the tools already on the wall

Instead of building your own bench, clamps and storage, you walk into a workshop where they are set up; you still need to know how to build, and the workshop has its own layout.

## Prototype fast, then decide

Dify is excellent for quick prototypes that stakeholders can try; decide later whether the app stays in Dify or moves to custom code.

**Quiz:** Which is part of Dify's feature set?

- [ ] A replacement for a SQL database
- [ ] Training foundation models from scratch
- [x] Visual workflows, knowledge bases, model provider connections and an API for each app
- [ ] A mobile operating system

*Answer:* Visual workflows, knowledge bases, model provider connections and an API for each app. Dify orchestrates models; it does not pretrain them.
