{"title":"Frequently asked questions (preview)","url":"https://jonas-paprotka.de/docs/dev-itpro/ai/ai-development-toolkit-faq/","by":"JOP-AL","license":"Copyright (c) Microsoft Corporation. CC BY 4.0, https://creativecommons.org/licenses/by/4.0/; code samples MIT. From https://learn.microsoft.com/en-us/dynamics365/business-central/dev-itpro/ai/ai-development-toolkit-faq, MicrosoftDocs/dynamics365smb-devitpro-pb at 828b6ee (2026-11-19). Changed: includes filled in, links to this site, and for agents shortened without dropping a fact.","credit":"Microsoft Corporation, CC BY 4.0","origin":"https://learn.microsoft.com/en-us/dynamics365/business-central/dev-itpro/ai/ai-development-toolkit-faq","source":{"name":"Business Central developer and IT pro docs","repo":"https://github.com/MicrosoftDocs/dynamics365smb-devitpro-pb","commit":"828b6ee72fcd75755627cd6011f3d20899217ab2","date":"2026-11-19"},"checked":"2026-11-19","description":"Get answers to common questions about designing agents in Business Central, including integration, AI development toolkit access, scheduling, instructions, and capabilities.","date":"2026-10-02","meta":{"topic":"faq"},"outline":[{"level":1,"title":"Frequently asked questions (preview)","slug":"frequently-asked-questions-preview","at":0,"chars":9965},{"level":2,"title":"Integration and connectivity","slug":"integration-and-connectivity","at":801,"chars":674},{"level":3,"title":"Can I link a mailbox to an agent like the Sales Order Agent?","slug":"can-i-link-a-mailbox-to-an-agent-like-the-sales-order-agent","at":833,"chars":642},{"level":2,"title":"Development","slug":"development","at":1475,"chars":460},{"level":3,"title":"Where can I find the AI development toolkit code?","slug":"where-can-i-find-the-ai-development-toolkit-code","at":1490,"chars":445},{"level":2,"title":"Scheduling and automation","slug":"scheduling-and-automation","at":1935,"chars":1001},{"level":3,"title":"How can I trigger an agent on a schedule (for example, daily)?","slug":"how-can-i-trigger-an-agent-on-a-schedule-for-example-daily","at":1964,"chars":972},{"level":2,"title":"Instructions and language","slug":"instructions-and-language","at":2936,"chars":2030},{"level":3,"title":"Can I write agent instructions in languages other than English?","slug":"can-i-write-agent-instructions-in-languages-other-than-english","at":2965,"chars":907},{"level":3,"title":"Is there a size limit for agent instructions?","slug":"is-there-a-size-limit-for-agent-instructions","at":3872,"chars":1094},{"level":4,"title":"Recommendations for effective instructions","slug":"recommendations-for-effective-instructions","at":4050,"chars":367},{"level":4,"title":"Signs your instructions might be too long","slug":"signs-your-instructions-might-be-too-long","at":4417,"chars":549},{"level":2,"title":"Choosing the right approach","slug":"choosing-the-right-approach","at":4966,"chars":1668},{"level":3,"title":"When should I use which tools?","slug":"when-should-i-use-which-tools","at":4997,"chars":1637},{"level":4,"title":"Design agents when","slug":"design-agents-when","at":5107,"chars":597},{"level":4,"title":"Use Copilot Studio agents when:","slug":"use-copilot-studio-agents-when","at":5704,"chars":437},{"level":4,"title":"Use AL code when:","slug":"use-al-code-when","at":6141,"chars":493},{"level":2,"title":"Agent capabilities","slug":"agent-capabilities","at":6634,"chars":957},{"level":3,"title":"Does the agent support filtering on pages?","slug":"does-the-agent-support-filtering-on-pages","at":6656,"chars":935},{"level":2,"title":"AI development toolkit app","slug":"ai-development-toolkit-app","at":7591,"chars":737},{"level":3,"title":"Do I need to install the AI development toolkit app to use the agent design experience?","slug":"do-i-need-to-install-the-ai-development-toolkit-app-to-use-the-agent-design-experience","at":7621,"chars":707},{"level":2,"title":"Provide feedback","slug":"provide-feedback","at":8328,"chars":871},{"level":3,"title":"How can I provide feedback for the agent design experience?","slug":"how-can-i-provide-feedback-for-the-agent-design-experience","at":8348,"chars":851},{"level":2,"title":"Related information","slug":"related-information","at":9199,"chars":766}],"text":"# Frequently asked questions (preview)\n[This article is prerelease documentation and is subject to change.]\n> IMPORTANT\n > This is a preview feature. Preview features might have restricted functionality and are subject to supplemental terms of use<https://www.microsoft.com/en-us/business-applications/legal/supp-powerplatform-preview/>.\n > The AI development toolkit is available in sandbox environments only. The Agent SDK for AL is available in sandbox environments and, starting from version 28.1, also in production environments.\n > This article is primarily intended for product owners, domain experts, consultants, and developers who want to experiment with AI capabilities in Business Central.\nThis article answers common questions about using the agent design experience in Business Central.\n## Integration and connectivity\n### Can I link a mailbox to an agent like the Sales Order Agent?\nYes. While there is no built-in configuration of email integration in the agent design experience, you can integrate email functionality with agents using the AI development toolkit. Learn more in the BCTech repo sample Agent and Email Integration Sample<https://github.com/microsoft/BCTech/pull/338/files>.\nThis allows you to:\n Process incoming emails automatically\n Trigger agent tasks based on email content\n Extract information from email attachments\n Send automated responses\nLearn more about triggering agents in Run an agent (preview)<ai-development-toolkit-run-agent>.\n## Development\n### Where can I find the AI development toolkit code?\nAccess the AI development toolkit at: BCApps/src/System Application/App/Agent<https://github.com/microsoft/BCApps/tree/main/src/System%20Application/App/Agent>\nThe toolkit provides:\n Agent Task Builder API\n Agent Task Message Builder API\n Custom Agent management codeunits\n Sample implementations and patterns\nSee Integrate with the Tasks AL API (preview)<ai-development-toolkit-tasks-api>.\n## Scheduling and automation\n### How can I trigger an agent on a schedule (for example, daily)?\nYou can schedule agent tasks using job queues in Business Central:\n1. Create a codeunit that uses the Agent Task Builder API to create tasks\n2. Set up a job queue entry to run the codeunit on your desired schedule\n3. Configure the job queue to run daily, hourly, or at custom intervals\nExample pattern:\n```al\ncodeunit 50100 \"Daily Agent Task Scheduler\"\n{\n    procedure CreateDailyTask()\n    var\n        AgentTaskBuilder: Codeunit \"Agent Task Builder\";\n        AgentTask: Record \"Agent Task\";\n    begin\n        AgentTask := AgentTaskBuilder\n            .Initialize(AgentUserSecurityId, 'Daily Review Task')\n            .SetExternalId('DAILY-' + Format(Today))\n            .AddTaskMessage('System', 'Perform daily review of pending items')\n            .Create();\n    end;\n}\n```\nLearn more about creating tasks programmatically in Integrate with the Tasks AL API (preview)<ai-development-toolkit-tasks-api>.\n## Instructions and language\n### Can I write agent instructions in languages other than English?\n> NOTE\n> While instructions can be in any language, consider that many Business Central field names, page names, and technical terms are in English. Mixing languages for technical references while keeping business logic in your local language often works well.\n> While other languages are supported, English instructions might yield the best results due to the training data used for the AI models.\nYou can write agent instructions in your local language, but bear the language considerations in mind. The AI models support multiple languages, allowing you to:\n Write instructions in the language most natural for your team\n Use domain-specific terminology in your native language\n Provide multilingual support for international implementations\nSee Write effective instructions for an agent (preview)<ai-development-toolkit-instructions>.\n### Is there a size limit for agent instructions?\nNo, there's no hard limit on instruction size. However, we recommend keeping instructions focused and concise for best results.\n#### Recommendations for effective instructions\n Optimal length\n  500-2000 words for most scenarios\n Focus on essentials\n  Include only relevant guidance and rules\n Use structure\n  Break instructions into clear sections (responsibilities, guidelines, steps)\n Avoid repetition\n  Don't repeat information that's already available in field tooltips or page descriptions\n#### Signs your instructions might be too long\n The agent frequently misses important steps\n Responses become inconsistent\n Processing time increases noticeably\n The agent struggles with context from earlier in the instructions\nIf your instructions exceed 3000-4000 words, consider:\n Splitting complex workflows into multiple specialized agents\n Moving static reference information to linked documentation\n Consolidating redundant or overlapping guidance\nSee Write effective instructions for an agent (preview)<ai-development-toolkit-instructions>.\n## Choosing the right approach\n### When should I use which tools?\nChoose the approach that best fits your scenario and stage of development:\n#### Design agents when\n Prototyping new AI-powered features before committing to development\n Testing AI feasibility for specific business scenarios\n Learning how AI can interact with Business Central data\n Demonstrating AI capabilities to stakeholders\n Rapid iteration is needed to refine requirements\n You're in a sandbox environment and want quick results\nLimitations:\nThese agents are for prototyping in sandbox environments. To deliver these agents to customers in production environments, use the AI development toolkit to productize adding integrations, setup etc. and ship in extensions.\n#### Use Copilot Studio agents when:\n Building production-ready conversational experiences\n Need integration with other products via the many supported Copilot Studio connectors\n Require advanced dialog management and conversation flows\n Want low-code development with visual designers\n Building customer-facing or employee-facing chat experiences\nExample scenarios:\nCustomer service chatbots, HR support agents, IT helpdesk automation.\n#### Use AL code when:\n Building custom Business Central extensions with AI features\n Need full control over the implementation and logic\n Require tight integration with Business Central objects and events\n Must meet specific performance or security requirements\n Building commercial solutions for Marketplace\n Need version control and ALM (Application Lifecycle Management)\nExample scenarios:\nCustom Copilot features, AI-powered document processing, and intelligent automation in extensions.\n## Agent capabilities\n### Does the agent support filtering on pages?\nYes, agents can filter data on pages. Filtering is one of the core capabilities that agents use to navigate and find specific records.\nSupported filtering operations:\n Filter by specific field values\n Apply multiple filter criteria\n Use comparison operators (equals, greater than, less than)\n Filter by date ranges\nAvailable agent capabilities:\nThe agent's ability to use filtering depends on:\n The profile assigned to the agent (which controls visible pages and fields)\n The permissions granted to the agent\n Instructions that guide when and how to filter\nExample instruction:\n```text\nTo find overdue invoices:\n1. Open the Sales Invoices page\n2. Filter the \"Due Date\" field to show dates before today\n3. Review the filtered list and prioritize by amount\n```\nLearn more about available keywords in Instruction keywords for an agent (preview)<ai-development-toolkit-instruction-keywords>.\n## AI development toolkit app\n### Do I need to install the AI development toolkit app to use the agent design experience?\nNo, the agent design experience is built directly into Business Central sandbox environments and shouldn't require a separate app installation. However, if you don't see the agent design experience, go to the Extension Management page, and ensure that the AI Development Toolkit extension is installed and enabled. If it's not installed, you can install it from Marketplace<https://marketplace.microsoft.com/product/PUBID.microsoftdynsmb%7CAID.business-central-agent-designer%7CPAPPID.f2d0130c-4b0e-437a-bcfd-0dc9f84ac961?tab=Overview>. Learn more about installing apps in /ui-extensions-install-uninstall#install.\n## Provide feedback\n### How can I provide feedback for the agent design experience?\nWe welcome your feedback to help us improve the agent design experience. In Business Central, there are several places where you can provide feedback on agents and the agent design experience. Use the Give feedback option here:\n Agent Task Log Entries page - provide feedback on specific agent tasks and their execution\n Agent Task Log page - share your thoughts on the overall agent performance and runtime behavior\nUse the thumbs up/down buttons to indicate whether the agent's response was helpful, and provide additional comments to help us understand your experience and improve the product here:\n Agent timeline- give feedback on the sequence of agent actions and interactions\n Edit instructions page - let us know about your experience writing and editing instructions for agents\n## Related information\nAI development toolkit overview (preview)<ai-development-toolkit-overview>\nCreate and activate an agent (preview)<ai-development-toolkit-agent-create>\nWrite effective instructions for an agent (preview)<ai-development-toolkit-instructions>\nIntegrate with the Tasks AL API (preview)<ai-development-toolkit-tasks-api>\nAttachment capabilities and limitations (preview)<ai-development-toolkit-attachments>\nSet up permissions and profiles (preview)<ai-development-toolkit-permissions-profiles>\nRun an agent (preview)<ai-development-toolkit-run-agent>\nBest practices for designing an agent (preview)<ai-development-toolkit-best-practices>\nTransparency note: Business Central AI development toolkit (preview)<transparency-note-ai-development-toolkit>"}