# The Dock & App drawer > The two panels you spend the most time in: the Dock for adding steps, the App drawer for configuring them. Two panels do most of the work on the canvas. The **Dock** along the bottom is where steps come from. The **App drawer** on the right is where you configure the one you have selected. ## The dock ![The Dock along the bottom of the canvas, with Run, Apps, Tools, AI, Subflows and Search](/images/docs/the-canvas/dock.webp) *Run sits on the left as its own control; the five groups to its right are where steps come from.* The Dock holds everything you can add to a workflow, in five groups: | Group | What is in it | | ------------ | --------------------------------------------------------------------------------------------------- | | **Apps** | The third-party integration library: both triggers and actions live here. | | **Tools** | Every built-in step, grouped by what it is for — see below. | | **AI** | Language-model steps such as AI Prompt and AI Agent. | | **Subflows** | Your other workflows, ready to call as a step. See [Subflow](/build/action-steps/trigger-workflow). | | **Search** | Find any app, trigger, or tool by name. | ### How Tools is organised The Tools tab groups the built-in steps by the job they do, with a line under each group name saying what belongs there: | Group | What it holds | | --------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | | **Start** | What sets the workflow running: [Scheduler](/build/triggers/scheduler), [Click to start](/build/triggers/click-to-start), [Webhook](/build/triggers/webhook) and [Chat Trigger](/build/triggers/chat). | | **Flow** | Splitting the path, repeating steps, waiting for a person: Conditions, Switch, Filter, Repeater, Wait, Human Review, Stop and Error, Do Nothing. | | **Data** | Everything that works on what the workflow carries, under four headings of its own. | Data is the one group with sub-headings: | Under Data | What it holds | | ------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Change** | Reshaping a record: Summarize, Change fields, Date, Custom Variables, AI Data Transform. | | **Lists** | Working on a list as a whole: Sort, Remove duplicates, Limit, Combine. | | **Convert** | Turning one format into another: Parse JSON, HTML to text, Split Text, plus the [Formatting and Helper Functions](/build/action-steps/helper-functions) collections. | | **Dev tools** | Talking to something outside Glow: HTTP Request, Code editor. | A search box under the groups finds any tool by name, so you never have to memorise the grouping. > **A trigger from a third-party app**, such as starting when a Google Doc is > created, lives under **Apps**, together with that app's actions. The **Start** > group in Tools holds the built-in triggers only. **Run** sits beside these rather than inside them: the button itself, plus a dropdown listing every step that can be the run point. See [Testing & Debugging](/build/the-canvas/testing-and-debugging). The [right-click context menu](/build/the-canvas/canvas-context-menu) offers the same groups wherever your cursor is. That is usually quicker once you know what you are looking for. ## The App drawer Click a step and the App drawer opens on the right. Most steps have three tabs: - **Setup:** the step's parameters. Which account it uses, which action it performs, and how its inputs are mapped. - **Executions:** what this step produced on previous runs. See [Step-level Executions](/build/core-concepts/executions). - **Test & Debug:** run this step on its own, edit its test data, and set its retry and error behaviour. See [Error Handling](/build/core-concepts/error-handling). The **Chat Trigger** replaces **Test & Debug** with **Chat**, where you send test messages and inspect the conversation. Flow-control steps that do not have independent test settings show **Test & Debug** as unavailable. > After a run, **Executions** is the fastest way to work out why the result was > not what you expected. It shows exactly what the step returned; what it > received is the step above's output, on that step's own Executions tab. ## Annotations The canvas is not only for steps. Three things can be dropped anywhere on it, none of which affect execution. They come from the **Sticky Note & Media** menu in the controls at the top right, or from the shortcut key: | Annotation | Shortcut | Use | | ----------------- | :----------: | -------------------------------------------------------------------------------- | | **Sticky note** | N | Comments, open questions, or an explanation of awkward logic. Supports Markdown. | | **Image** | I | Screenshots, architecture diagrams, reference designs. | | **YouTube video** | Y | A walkthrough or recording, so context stays with the workflow. | These are what make a workflow readable to whoever opens it next, including you, months later. Everyone on the team sees them live, as covered in [Real-Time Editing](/build/the-canvas/real-time-editing). ## What's Next? - Learn what a step is made of in [Steps and Connections](/build/core-concepts/steps-and-the-canvas). - Reach the same menus faster with the [Context Menu](/build/the-canvas/canvas-context-menu).