{"revision":"c10276e953456835cdbb1f3a60d58ae1980be353","document":{"id":2153,"path":"application-development/exploring-autonomous-engineer.md","title":"Exploring Autonomous Engineer","bundle":null,"doc_type":null,"release":"australia","body":"\n# Exploring Autonomous Engineer\n\nAutonomous Engineer is an agentic worker that accepts requirements, generates a structured implementation plan, and builds all work items in parallel using background agents.\n\n## Autonomous Engineer overview\n\nImplementing applications on the ServiceNow AI Platform can require significant manual effort between defining requirements and delivering value to users. Autonomous Engineer reduces that time to value by taking your specifications as input and driving the full implementation lifecycle end-to-end.\n\nAutonomous Engineer follows two sequential phases: planning and execution. You remain in control throughout both phases. Autonomous Engineer requires your review and approval before building.\n\nAutonomous Engineer is available in ServiceNow Studio and supports custom applications. Autonomous Engineer uses agent packs with product-specific domain knowledge to run implementations.\n\n## How Autonomous Engineer works\n\n-   **Planning phase**\n\n    Provide your requirements in any of the following forms:\n\n    -   A natural language prompt or paragraph.\n    -   A file upload, such as a CSV of user stories.\n    Autonomous Engineer reads your requirements and conducts an interview to resolve ambiguous or incomplete details. It queries your instance to identify existing artifacts and metadata such as tables, roles, fields, and catalog items, so existing artifacts aren't duplicated.\n\n    The output of planning is a plan with work items. Work items are structured in an Agile user story format and include a description and acceptance criteria. They are grouped into logical sections and sequenced using a dependency graph so that each work item executes only after its dependencies are complete.\n\n    Review every work item before execution begins. You can read the details, edit the content, and make changes to align the plan with your requirements. Nothing is built until you approve the plan and its work items.\n\n-   **Execution phase**\n\n    After you approve the plan, Autonomous Engineer generates a background agent for each work item and builds all work items in parallel. Each agent runs in isolation. The plan dashboard displays the status of each work item as it progresses.\n\n    For each work item, Autonomous Engineer builds the required metadata, runs Test Agent to generate Automated Test Framework \\(ATF\\) tests, and runs those tests. If tests fail, Autonomous Engineer identifies the root cause and attempts to resolve the failure through several retry cycles. Work items that can't be resolved automatically appear in the plan dashboard and the chat panel for your review.\n\n    When all work items are complete, Autonomous Engineer generates an update set for the plan. Export the update set and deploy it for user acceptance testing or to a production environment.\n\n    To view and manage all your plans across sessions, use the **Plans** view in the Build Agent navigator panel. For more information, see [Plans in Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-plans.md).\n\n\n## Manual steps\n\nSome implementation tasks fall outside what Autonomous Engineer can perform autonomously, such as activating plugins. When a plan includes steps that require manual action, Autonomous Engineer lists those steps in a manual verification section before you run the plan. Completing manual steps before execution reduces the risk of work items failing because of missing dependencies.\n\n## Autonomous Engineer benefits\n\nAutonomous Engineer accelerates application development on the ServiceNow AI Platform by enabling:\n\n-   Parallel execution of work items, reducing total implementation time.\n-   Automatic ATF test generation and self-healing retry cycles.\n-   Structured planning with dependency sequencing so artifacts are built in the correct order.\n-   A single update set output, ready for deployment to test or production environments.\n-   Consolidation of requirements across multiple sources and systems.\n-   Clarification and validation of complex requirements.\n\n## What to explore next\n\nFor more information about configuring and using Autonomous Engineer, see:\n\n-   [Configure Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/configure-autonomous-engineer.md)\n-   [Using Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-using-autonomous-engineer.md)\n-   [Autonomous Engineer reference](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/autonomous-engineer-reference-landing.md)\n\n-   **[Autonomous Engineer and Build Agent](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-build-agent-relationship.md)**  \nAutonomous Engineer and Build Agent are separate products that work together. Autonomous Engineer uses Build Agent as its execution layer.\n-   **[Agent packs for Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown)**  \nAgent packs bundle the domain knowledge, tools, and skills that Autonomous Engineer uses to implement ServiceNow products.\n-   **[Autonomous Engineer chat panel](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-chat-panel.md)**  \nThe Autonomous Engineer chat panel within Build Agent is where you interact with the AI agent during planning and development. Use it to submit requests, review responses, run tests, and apply generated code.\n-   **[Using the dashboard](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-plan-dashboard.md)**  \nThe Autonomous Engineer dashboard displays the status of every work item in your plan and provides navigation to individual work items, milestones, and pre-flight verification items.\n-   **[Resilience in Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-resiliency.md)**  \nAutonomous Engineer monitors background agents during execution and automatically detects and retries work items that get stuck or unresponsive.\n-   **[Supported models for Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-supported-models.md)**  \nLearn which AI models and versions Autonomous Engineer supports and how to change them. Use this information to verify compatibility and select the right model for your task.\n-   **[Supported tools for Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-supported-tools.md)**  \nAutonomous Engineer tools support application development tasks such as semantic search, schema inspection, code search, planning, UI validation, database querying, app navigation, and script execution. Each tool extends what Autonomous Engineer can do during a build session.\n-   **[MCP connections and Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-mcp-connections.md)**  \nMCP connections enable Autonomous Engineer to access external tools and resources through standardized communication, extending the planning and implementation workflow with data and context from third-party systems.\n-   **[Playbooks in Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-playbooks.md)**  \nUse Autonomous Engineer to author and manage Playbook Designer artifacts through a conversation. You can generate playbook structures, configure activities, set runtime permissions, and define launcher configurations without manually navigating the Playbook Designer UI.\n-   **[Autonomous Engineer governance](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-governance.md)**  \nGovernance controls for Autonomous Engineer are managed through Build Agent.\n-   **[Autonomous Engineer limitations](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/ae-limitations.md)**  \nLimitations that apply to Build Agent also apply to Autonomous Engineer.\n\n**Parent Topic:**[Autonomous Engineer](https://raw.githubusercontent.com/ServiceNow/ServiceNowDocs/australia/markdown/australia/application-development/autonomous-engineer.md)\n\n"}}