--- title: Create and Modify Playbooks with Hero AI slug: turbine-user-guide/create-and-modify-playbooks-with-hero-ai docTags: createdAt: 2026-05-29T09:50:25.563Z --- Use **Playbook Building Mode** (also called **Text to Playbook**) to create or change playbook flows on the canvas with natural language. The **Playbook Generator Agent** applies your prompts to the selected flow or flows. Starting in **Turbine 26.2.0**, **Playbook Builder v2** added **clarifying questions** before building and **progress detail** messages while generation runs. Starting in **Turbine 26.3.0**, the agent can **edit multiple flows in one request**—select flows on the canvas or name them in natural language, then review and **Accept all** or **Reject all** proposed changes. See [Edit Multiple Flows in One Request](docId\:TmMNjGlC0e9lIPO37KZPB). You can still edit a **single flow** by selecting it on the canvas or naming it in your prompt. For manual drag-and-drop creation, see [How to Create a Playbook](docId:0-KvG5p4LTHw75ItmpUWp). For supported operations, limits, and trigger rules, see [Playbook Generator Agent Reference](docId:3JkvN5yplM_U06tfjHiAC). ## Choose Your Path | I want to… | Start here | | -------------------------------------------------- | ----------------------------------------------------------------------------------- | | Create the first flow in a new playbook with AI | [Create a New Flow with Hero AI](docId\:TmMNjGlC0e9lIPO37KZPB) | | Change an existing flow on the canvas | [Modify an Existing Flow](docId\:TmMNjGlC0e9lIPO37KZPB) | | Edit several flows in one Hero AI prompt | [Edit Multiple Flows in One Request](docId\:TmMNjGlC0e9lIPO37KZPB) | | Accept or reject AI changes across flows | [Review and Accept Hero AI Changes](docId\:TmMNjGlC0e9lIPO37KZPB) | | Understand what a flow does without editing it | [Ask Questions About Existing Flows](docId\:TmMNjGlC0e9lIPO37KZPB) | | Answer agent questions before a build completes | [Clarifying Questions](docId\:TmMNjGlC0e9lIPO37KZPB) (playbook or component canvas) | | See status messages during generation | [Progress During Playbook Generation](docId\:TmMNjGlC0e9lIPO37KZPB) | | See what the agent supports (including multi-flow) | [Playbook Generator Agent Reference](docId:3JkvN5yplM_U06tfjHiAC) | | Confirm I am in the right Hero AI mode | [Enter Playbook Building Mode](docId\:TmMNjGlC0e9lIPO37KZPB) | | Understand container vs flow when saving | [Playbooks Overview](docId\:Dp-AGNnnE2MJJFGtV2LZ5) | ## Enter Playbook Building Mode There is **no separate Building Mode toggle** in settings. The companion enters **Playbook Building Mode** when Hero AI opens **with canvas context** (the product knows which playbook and flow you are editing). | How you open Hero AI | Mode | Typical use | | ------------------------------------------------------------------------- | -------------------------- | ------------------------------------------------------------------------------------------------------------ | | **Hero AI** from the main application toolbar (outside the canvas editor) | General companion | Questions and chat not tied to a specific canvas flow; see [Hero AI Companion](docId\:crQKFVnGpCz_wJ8xtHDs7) | | **Build with Hero AI** on the canvas toolbar | **Playbook Building Mode** | Generate or change flows with the Playbook Generator Agent | | **Start building with Hero AI** on an empty canvas | **Playbook Building Mode** | Create the first flow in a new playbook | | **Hero AI Companion** while a playbook canvas is already open | **Playbook Building Mode** | Modify or explain the selected flow | | Navigate away from the playbook canvas editor (panel can stay open) | General companion | Building Mode ends; see [Hero AI Companion](docId\:crQKFVnGpCz_wJ8xtHDs7) | ### Steps to Enter Playbook Building Mode 1. Open a playbook in the canvas editor (see [How to Create a Playbook](docId:0-KvG5p4LTHw75ItmpUWp) to create one manually first, if needed). 2. Do one of the following: - Click **Build with Hero AI** on the canvas toolbar, or - On an empty canvas, click **Start building with Hero AI**, or - Open the **Hero AI Companion** from the playbook canvas. 3. Confirm that the companion shows **Playbook Building Mode** (see below) before you send a build or modify prompt. ### What You Should See in Playbook Building Mode When Building Mode is active, the Hero AI companion typically shows: - A banner such as **Entered Playbook Building Mode** or **Playbook Building Mode** near the top of the panel. :::hint{type="info"} If you open Hero AI only from the main toolbar and do not see Building Mode, you are in the general companion. Open the playbook canvas and use **Build with Hero AI** or **Start building with Hero AI** instead. ::: To stop generation while a playbook is being built, click **Cancel** on the canvas progress overlay or in the companion when shown. For enable, disable, delete, and trigger rules, see [Playbook Generator Agent Reference](docId:3JkvN5yplM_U06tfjHiAC). ## Edit Multiple Flows in One Request Starting in **Turbine 26.3.0**, **Playbook Builder V3** can apply coordinated edits across **more than one flow** in a single Hero AI session. ### How to Target Multiple Flows 1. Open a multi-flow playbook in the canvas editor and enter **Playbook Building Mode**. 2. Do one or both of the following: - **Canvas:** Select a flow on the canvas. Unless you name other flows in the prompt, the agent scopes edits to that selection. - **Natural language:** Name flows by their **titles** (for example: "In **Enrich Alerts** and **Notify SOC**, add a VirusTotal check after the first action."). Prefer titles over canvas position words such as "the first flow" or "the leftmost flow"—the agent asks for a title instead of guessing position. 3. Send the prompt. Answer any **clarifying questions** if the agent asks which flows or actions you mean. 4. When generation finishes, review the **in-chat flow summary** (changes grouped by flow with Added / Removed / Modified lines) and the **canvas summary** footer — see [Review and Accept Hero AI Changes](docId\:TmMNjGlC0e9lIPO37KZPB). 5. **Accept all** or **Reject all**, or accept/reject per flow, then **Save** the playbook. ### Example Multi-Flow Prompts - "Add the same Slack notification step to **Triage** and **Escalate**." - "Disable **Legacy Ingest** and add a schedule every 15 minutes to **Notify SOC**." - "Create a new flow called **Archive Closed** that runs daily at 02:00 UTC and archives closed cases." - "Enable all flows." / "Disable all flows." (agent updates each flow’s enabled state) ### What Multi-Flow Editing Does Not Change - You still **Save** the playbook after accepting changes so container and flow layers persist. - **Record event** and **playbook button** triggers remain limited (filter existing only) — see [Playbook Generator Agent Reference](docId:3JkvN5yplM_U06tfjHiAC). - The agent **cannot** connect an action in one flow to an action in another flow (cross-flow edges are not supported). - Content locking and concurrent-tab rules still apply. :::hint{type="info"} If the same playbook is open in more than one browser tab or window, saves can overwrite each other. See [Playbook Generator Agent Reference](docId:3JkvN5yplM_U06tfjHiAC) in **Playbook Generator Agent Reference**. ::: ## Review and Accept Hero AI Changes When Hero AI proposes canvas changes across one or more flows, Turbine shows two related summaries: ### Canvas Summary (footer) A footer panel appears with a title such as **Hero AI made changes to 2 flows** (or **added** / **removed** flow phrasing). For each affected flow it lists **Added**, **Removed**, and **Modified** items. You can: | Action | What it does | | -------------------------------- | --------------------------------------------------------------------------------------------------------- | | **Accept all** | Keeps all pending Hero AI changes across the listed flows | | **Reject all** | Discards all pending Hero AI changes (confirm when prompted) | | Per-flow **Accept** / **Reject** | Accept or reject changes for one flow (tooltips: **Accept changes to flow** / **Reject changes to flow**) | | Click a change label | Focuses that action or flow on the canvas (removed items are not focusable) | Use the status labels to decide what to review: | Status | When it appears | | ------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Added** | A new action or flow that Hero AI created for this prompt | | **Removed** | An action or flow that Hero AI removed for this prompt | | **Modified** | A meaningful change to that action, such as configuration updates or a change to the outgoing edge type (for example, **On Success** to **On Failure**) | :::hint{type="info"} When Hero AI only **adds** or **removes** an action, the parent action is **not** listed as **Modified** solely because its outgoing connection was rewired to the new or next action (same edge type). For example, a prompt such as "Add a logging note component at the end of every flow" should mainly show **Added** items, not a long list of unchanged parents marked **Modified**. Review **Modified** items when the summary shows a real config or edge-type change on that action. ::: ### In-Chat Flow Summary The companion message for that turn can include a collapsible summary **by flow** with the same Added / Removed / Modified structure. Use it to understand what changed for that prompt; bulk accept/reject remains on the canvas summary. :::hint{type="info"} Accepting changes updates the canvas draft. **Save** the playbook to persist. If you reject changes and prompt again for the same flow, the agent rebuilds from the current accepted state. ::: ## Create a New Flow with Hero AI 1. Open a playbook in the canvas editor and enter **Playbook Building Mode**. 2. On an empty canvas, click **Start building with Hero AI**. 3. Describe what you want to build (for example: "Create a playbook that fetches alerts from CrowdStrike every 5 minutes, extracts observables, and enriches them with VirusTotal."). Each new flow must include at least one trigger or action (empty flows are not supported). 4. Answer any **clarifying questions** if the agent asks — see [Clarifying Questions](docId\:TmMNjGlC0e9lIPO37KZPB). 5. Review **progress detail** messages and the generated flow — see [Progress During Playbook Generation](docId\:TmMNjGlC0e9lIPO37KZPB). When the canvas summary appears, also review the [canvas and in-chat summaries](docId\:TmMNjGlC0e9lIPO37KZPB). 6. Iterate with follow-up prompts. 7. **Save** the playbook when you are satisfied so both the flow and container layers persist. ## Modify an Existing Flow 1. Enter **Playbook Building Mode** on the playbook canvas. 2. Select the flow you want to change on the canvas, name it in the prompt, or @-mention the flow in chat. 3. Ask Hero AI to update it (for example: "Add an IP reputation check after the enrichment step and branch if malicious."). 4. Answer clarifying questions if prompted. 5. Review progress messages and the updated flow. Use [Review and Accept Hero AI Changes](docId\:TmMNjGlC0e9lIPO37KZPB) when the canvas summary appears. 6. **Save** the playbook. ## Ask Questions About Existing Flows The Playbook Generator Agent is not limited to creating or modifying flows. You can also use it to understand what an existing flow does. Open the **Hero AI Companion** from a playbook canvas in Building Mode and ask questions such as: - "What does this flow do?" - "Explain the steps in this playbook." - "Why would this flow branch at the condition step?" Hero AI analyzes the selected flow and provides a plain-language explanation, which helps onboard new team members, audit automation logic, or troubleshoot unexpected behavior. ## Clarifying Questions Starting in **Turbine 26.2.0**, both **Playbook Building Mode** and **Component Building Mode** may pause and ask **clarifying questions** when your prompt needs more detail before it can build or change a flow or component safely. **When they appear:** - After you submit a build or modify prompt in **Playbook Building Mode** or **Component Building Mode** - When the agent needs choices such as trigger type, target system, error handling, or branch behavior **What you see:** 1. The Hero AI panel shows a brief message that it needs more detail before it can continue building. 2. An optional intro message from the agent explains why it paused. 3. One question at a time with **Question X of Y** progress. 4. An instruction line: **Select one option below** (single-select) or **Select one or more options below** (multi-select). 5. Selectable options. Options with descriptions show an info icon; hover for the full description. 6. **Back**, **Next**, and **Save** buttons to move through questions and submit answers. **How to respond:** | Step | Action | | ---- | ------------------------------------------------------------------------------------------------------------------------------ | | 1 | Read the question and select one or more options | | 2 | Click **Next** to continue, or **Save** on the last question | | 3 | Use **Back** to change a prior answer | | 4 | If **Other** is available, select it and enter custom text in the main chat input field (required before **Next** or **Save**) | After you click **Save**, your answers are submitted to the agent and generation continues. A summary of your clarifying answers may appear in the chat history. :::hint{type="info"} You cannot send a new build prompt while clarifying questions are active. Complete or navigate away from the question flow before starting a different request. While clarifying questions are open, the canvas progress overlay may show **Waiting for user response** with an alert icon instead of the generation sparkle. ::: :::hint{type="info"} Answer clarifying questions with the integration and trigger you actually use in production. Vague answers produce generic flows that need more manual edits. ::: ## Progress During Playbook Generation Starting in **Turbine 26.2.0**, **Text to Playbook** shows **progress detail** status messages on the canvas and in the Hero AI companion while the Playbook Generator Agent works. **Where messages appear:** | Location | What you see | | --------------------------- | -------------------------------------------------------------------------------------------------------------------- | | **Canvas progress overlay** | Up to three rolling status lines with animated dots on the current line and a **Cancel** button while the agent runs | | **Hero AI companion chat** | A live progress stack under the assistant message for the current run | **What messages mean:** | Message pattern (examples) | What the agent is doing | | -------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- | | **Processing prompt** | The agent accepted your prompt and is starting the run | | Reading **playbook** or **trigger schemas** | Loading current flow structure and trigger configuration | | Reading **native action schemas** or **component** | Discovering available actions and connectors | | **Building playbook** or updating flow steps | Adding or changing actions on the canvas | | **Thinking** | A detailed in-progress update is underway (canvas overlay only when the underlying hub message exceeds roughly 75 characters) | | **Waiting for user response** | The agent is waiting for clarifying-question answers in the companion | | A short custom sentence | The agent supplied a specific description for the current tool step | Messages update in real time. Wording may vary between runs. Long hub messages may display as **Thinking** on the canvas overlay while the full detail remains in the companion progress stack. **While generation is running:** 1. Canvas editing and some companion actions are limited until the agent finishes, you cancel, or clarifying questions appear. 2. Click **Cancel** on the canvas overlay to stop the current generation attempt. 3. Do not navigate away from the playbook editor if you need to monitor completion. **After generation completes:** - Review the flow on the canvas before saving. - Use the companion to ask follow-up modify prompts or clarifying questions on the next iteration. ## Prompt Tips - Be explicit about triggers, actions, and data sources. - Use step placement language (for example, "after \[step name]" or "before \[step name]"). - Ask for error handling when external integrations are involved. ## See Also - [Hero AI](docId\:swZ8uETVmVoOJwGm_xNK-) — overview of all Hero AI features and building modes - [Playbook Generator Agent Reference](docId:3JkvN5yplM_U06tfjHiAC) — supported scenarios, multi-flow matrix, limits, and validation notes - [Organizing Playbook in the Canvas](docId\:ds_geaoGH--2ujXg-ROYS) — canvas tools (multi-select, Group, cut, copy, paste, and zoom) - [Playbooks Overview](docId\:Dp-AGNnnE2MJJFGtV2LZ5) — canvas layout, flows, and data model - [Hero AI Companion](docId\:crQKFVnGpCz_wJ8xtHDs7) — general companion (records, reports, tenant questions) - [Test Console](docId:1D0fiftjsdox88PVpK5fS) — validate flows after AI changes