---
id: video/MNwTt06ZxwY
type: video
title: "Microsoft presents: Building and shipping agents in Business Central"
summary: "Building, testing and shipping agents in Business Central: creating and configuring agents in the client (27.4 and newer), model selection in version 28, task simulation, exporting definitions, packaging agents as extensions for production, the AL agent metadata interfaces, and AI evaluation tools for tests."
tier: community
language: en
tags:
  - agents
  - agent configuration
  - permissions
  - profiles
  - instructions
  - task simulation
  - testing
  - agent models
  - export
  - user access
  - task logging
  - agent deployment
system: copilot
review:
  state: reviewed
  by: opus
  at: "2026-10-07T23:17:44.510Z"
  flags: []
generated:
  at: "2026-10-09T00:29:02.995Z"
  pipeline: 0.2.0
  prompts:
    extract-video: 1
    summarize-video: 2
  input_hash: 0ad2d85b47ac2e83c01a0781fe523eaf9cc6bf39a37cfdaf27b0a7514e51111b
evidence:
  - kind: video
    url: https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1430s
    title: "Optional interface methods: announced"
    date: "2026-10-01T05:38:51.000Z"
    commit: null
    t: 1430
    quote: This is a feature that's coming up. Actually, Quintin here worked on it, and he's a he's ready to release it almost
  - kind: video
    url: https://www.youtube.com/watch?v=MNwTt06ZxwY&t=4s
    title: "Microsoft presents: Building and shipping agents in Business Central"
    date: "2026-10-01T05:38:51.000Z"
    commit: null
    t: 4
    quote: Agents are no longer just experiments. They're becoming a first-party product inside of Business Central.
  - kind: video
    url: https://www.youtube.com/watch?v=MNwTt06ZxwY&t=90s
    title: "Microsoft presents: Building and shipping agents in Business Central"
    date: "2026-10-01T05:38:51.000Z"
    commit: null
    t: 90
    quote: This is all available for you today. You can look at it in the client. Um from 27.4 and newer.
  - kind: video
    url: https://www.youtube.com/watch?v=MNwTt06ZxwY&t=157s
    title: "Microsoft presents: Building and shipping agents in Business Central"
    date: "2026-10-01T05:38:51.000Z"
    commit: null
    t: 157
    quote: Maybe this instance of my agent needs to work in a certain locale. Maybe it's a US decimal points or it's a European decimal
  - kind: video
    url: https://www.youtube.com/watch?v=MNwTt06ZxwY&t=256s
    title: "Microsoft presents: Building and shipping agents in Business Central"
    date: "2026-10-01T05:38:51.000Z"
    commit: null
    t: 256
    quote: As you mature the agent, you need to come back and figure out if the agent is seeing too much or not seeing enough
  - kind: video
    url: https://www.youtube.com/watch?v=MNwTt06ZxwY&t=382s
    title: "Microsoft presents: Building and shipping agents in Business Central"
    date: "2026-10-01T05:38:51.000Z"
    commit: null
    t: 382
    quote: you wouldn't even see this agent models list. That would be for 28 where you have now this drop down
links:
  learn: []
  objects:
    - object/codeunit/4321
    - object/page/4316
  features:
    - feature/573352
  topics: []
  localizations: []
  videos: []
  posts: []
  guidelines: []
video_id: MNwTt06ZxwY
channel: yt-mibuso
source_name: mibuso.com / BC TechDays
url: https://www.youtube.com/watch?v=MNwTt06ZxwY
published_at: "2026-10-01T05:38:51.000Z"
duration_s: 2752
captions: derived
audience:
  - developer
  - functional consultant
  - administrator
  - partner
chapters:
  - t: 0
    title: Introduction to agents as a first-party product
  - t: 111
    title: "Agent configuration: name, language, region, and user access"
  - t: 242
    title: Profiles, permissions, and instructions - iterative improvement
  - t: 369
    title: Model selection and agent configuration options
  - t: 428
    title: Testing agents with task simulation and templates
  - t: 543
    title: "Task execution analysis: agent actions and user interventions"
  - t: 764
    title: Exporting and managing agent definitions
  - t: 928
    title: Moving agents to production via extensions
  - t: 1000
    title: Agent template and project structure
  - t: 1083
    title: Copilot capabilities and agent lifecycle management
  - t: 1159
    title: Agent metadata provider and factory interface
  - t: 1352
    title: Agent metadata runtime configuration and optional interface methods
  - t: 1473
    title: Evolving agents for production complexity and user intervention
  - t: 1614
    title: AI evaluation tools, testing, and quality tracking
features:
  - name: Agent creation and configuration
    status: unclear
    t: 111
    verified: false
    status_source: video
  - name: Manage user access for agents
    status: unclear
    t: 180
    verified: false
    status_source: video
  - name: Profiles and permissions configuration
    status: unclear
    t: 242
    verified: false
    status_source: video
  - name: Agent instructions configuration
    status: unclear
    t: 312
    verified: false
    status_source: video
  - name: Agent model selection
    status: unclear
    t: 369
    verified: false
    status_source: video
  - name: Agent task simulation and testing
    status: unclear
    t: 428
    verified: false
    status_source: video
  - name: Task templates
    status: unclear
    t: 487
    verified: false
    status_source: video
  - name: Task log and execution tracking
    status: unclear
    t: 533
    verified: false
    status_source: video
  - name: Agent tools and capabilities
    status: unclear
    t: 626
    verified: false
    status_source: video
  - name: User intervention in agent workflows
    status: unclear
    t: 704
    verified: false
    status_source: video
  - name: Download agent instructions
    status: unclear
    t: 775
    verified: false
    status_source: video
  - name: Export agent definition
    status: unclear
    t: 803
    verified: false
    status_source: video
  - name: Production agent deployment via extensions
    status: unclear
    t: 945
    verified: false
    status_source: video
  - name: Agent template for AL projects
    status: unclear
    t: 1000
    verified: false
    status_source: video
  - name: Copilot capability feature gate
    status: unclear
    t: 1083
    verified: false
    status_source: video
  - name: Agent metadata provider enumeration
    status: unclear
    t: 1159
    verified: false
    status_source: video
  - name: Agent factory interface
    status: unclear
    t: 1217
    verified: false
    status_source: video
  - name: Agent metadata interface for runtime configuration
    status: unclear
    t: 1352
    verified: false
    status_source: video
  - name: Take agent task execution interface
    status: unclear
    t: 1398
    verified: false
    status_source: video
  - name: Optional interface methods
    status: ga
    t: 1430
    verified: true
    status_source: roadmap
    roadmap_ids:
      - "573352"
  - name: Complex agent scenarios with user intervention
    status: unclear
    t: 1473
    verified: false
    status_source: video
  - name: AI evaluation tools for agents
    status: unclear
    t: 1614
    verified: false
    status_source: video
  - name: Copilot credit consumption visibility in tests
    status: unclear
    t: 1649
    verified: false
    status_source: video
  - name: Agent instance experimentation
    status: unclear
    t: 1662
    verified: false
    status_source: video
  - name: AI evals suite page
    status: unclear
    t: 1687
    verified: false
    status_source: video
  - name: Budget controls for agent tests
    status: unclear
    t: 1814
    verified: false
    status_source: video
  - name: Agent task drill-down and troubleshooting
    status: unclear
    t: 1882
    verified: false
    status_source: video
  - name: Copilot credit consumption visibility for non-admins
    status: unclear
    t: 2013
    verified: false
    status_source: video
  - name: Test run history and trend tracking
    status: unclear
    t: 2043
    verified: false
    status_source: video
  - name: Turn-based test design pattern
    status: unclear
    t: 2067
    verified: false
    status_source: video
  - name: AI development toolkit evaluation extension
    status: unclear
    t: 2179
    verified: false
    status_source: video
  - name: Agent test library
    status: unclear
    t: 2202
    verified: false
    status_source: video
  - name: Data-driven agent test approach
    status: unclear
    t: 2151
    verified: false
    status_source: video
  - name: AI test suite XML configuration standard
    status: unclear
    t: 2316
    verified: false
    status_source: video
  - name: AI test type code unit
    status: unclear
    t: 2398
    verified: false
    status_source: video
  - name: AI test type property
    status: unclear
    t: 2356
    verified: false
    status_source: video
  - name: Agent test context code unit
    status: unclear
    t: 2442
    verified: false
    status_source: video
  - name: YAML-based test data sets
    status: unclear
    t: 2500
    verified: false
    status_source: video
  - name: Run turn and wait method
    status: unclear
    t: 2579
    verified: false
    status_source: video
  - name: Data-driven agent test framework
    status: unclear
    t: 2641
    verified: false
    status_source: video
  - name: Agent debugging with break on next
    status: unclear
    t: 2699
    verified: false
    status_source: video
objects_mentioned:
  - other agent
  - other agent models list
  - other task template
  - other task log
  - other agents list
  - enum agent metadata provider
  - interface factory
  - interface agent metadata
  - interface take agent task execution
  - page AI evals suite pages
  - codeunit agent test context
  - codeunit AI test context
  - codeunit agent
  - page agent list
quotes:
  - t: 4
    text: Agents are no longer just experiments. They're becoming a first-party product inside of Business Central.
    check: exact
  - t: 90
    text: This is all available for you today. You can look at it in the client. Um from 27.4 and newer.
    check: exact
  - t: 157
    text: Maybe this instance of my agent needs to work in a certain locale. Maybe it's a US decimal points or it's a European decimal
    check: exact
  - t: 256
    text: As you mature the agent, you need to come back and figure out if the agent is seeing too much or not seeing enough
    check: exact
  - t: 382
    text: you wouldn't even see this agent models list. That would be for 28 where you have now this drop down
    check: exact
---

# Microsoft presents: Building and shipping agents in Business Central

> Building, testing and shipping agents in Business Central: creating and configuring agents in the client (27.4 and newer), model selection in version 28, task simulation, exporting definitions, packaging agents as extensions for production, the AL agent metadata interfaces, and AI evaluation tools for tests.

[Watch on YouTube](https://www.youtube.com/watch?v=MNwTt06ZxwY) · mibuso.com / BC TechDays · 2026-10-01 · 45:52 · tier community · reviewed (checked by Opus)

## Overview

The session from BC TechDays presents agents as a first-party product in Business Central. It walks through creating an agent instance in the client, setting name, language, region and user access, and iterating on profile, permissions and instructions. It then shows testing with task simulation, task templates and the task log, including user intervention steps.

The second half covers shipping. Agents created with instructions in the client cannot run in production, so they are exported and packaged as an extension, with the speaker saying production is supported from version 28. The AL side covers the agent metadata provider enum, the factory and metadata interfaces, and the Copilot capability as a feature gate. The session closes with AI evaluation tools: data-driven, turn-based tests, YAML data sets, an agent test library, credit budgets and run history, available in sandbox only.

## Key points

- Agents created with instructions in the client cannot run in production. They must be packaged as an extension and published to the production environment; the speaker says production is supported from version 28.
- Model choice is a dropdown in version 28 (GPT-4, GPT-4.1, with GPT-5 newer). In 27.4 the agent models list is not shown. Models differ in cost, and the speaker expects availability to grow as more models are validated.
- Test agents with task simulation (for email: subject, from, message, attachments) and reusable task templates. The task log shows each run's steps, tools used, and user intervention requests.
- Export the agent definition from the agents list to collect profile, permissions, instructions, locale and version. Instructions can also be downloaded on their own for an extension.
- Disabling the Copilot capability deactivates all agent instances that use it. Re-enabling it restores them.
- AI evaluation tools run in sandbox only, use your own data sets, and can feed CI/CD. Agent tests are turn-based, use YAML data sets and an AI test type code unit, and depend on the AI development toolkit evaluation extension.
- Budget limits exist at environment and company level. They block new tasks but let running tasks finish, so spend can slightly exceed the limit. Test results show Copilot credits consumed per task; the total tokens column shows only test framework tokens (such as LLM as judge calls), not the agents' own tokens.

## Chapters

- [0:00](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=0s) Introduction to agents as a first-party product
- [1:51](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=111s) Agent configuration: name, language, region, and user access
- [4:02](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=242s) Profiles, permissions, and instructions - iterative improvement
- [6:09](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=369s) Model selection and agent configuration options
- [7:08](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=428s) Testing agents with task simulation and templates
- [9:03](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=543s) Task execution analysis: agent actions and user interventions
- [12:44](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=764s) Exporting and managing agent definitions
- [15:28](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=928s) Moving agents to production via extensions
- [16:40](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1000s) Agent template and project structure
- [18:03](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1083s) Copilot capabilities and agent lifecycle management
- [19:19](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1159s) Agent metadata provider and factory interface
- [22:32](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1352s) Agent metadata runtime configuration and optional interface methods
- [24:33](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1473s) Evolving agents for production complexity and user intervention
- [26:54](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1614s) AI evaluation tools, testing, and quality tracking

## Features

| Feature | Status | At | Evidence |
|---|---|---|---|
| Agent creation and configuration | status not stated, demoed | [1:51](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=111s) |  |
| Manage user access for agents | status not stated, demoed | [3:00](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=180s) |  |
| Profiles and permissions configuration | status not stated, demoed | [4:02](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=242s) |  |
| Agent instructions configuration | status not stated, demoed | [5:12](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=312s) |  |
| Agent model selection | status not stated, demoed | [6:09](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=369s) |  |
| Agent task simulation and testing | status not stated, demoed | [7:08](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=428s) |  |
| Task templates | status not stated, demoed | [8:07](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=487s) |  |
| Task log and execution tracking | status not stated, demoed | [8:53](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=533s) |  |
| Agent tools and capabilities | status not stated, demoed | [10:26](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=626s) |  |
| User intervention in agent workflows | status not stated, demoed | [11:44](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=704s) |  |
| Download agent instructions | status not stated, demoed | [12:55](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=775s) |  |
| Export agent definition | status not stated, demoed | [13:23](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=803s) |  |
| Production agent deployment via extensions | status not stated | [15:45](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=945s) |  |
| Agent template for AL projects | status not stated, demoed | [16:40](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1000s) |  |
| Copilot capability feature gate | status not stated, demoed | [18:03](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1083s) |  |
| Agent metadata provider enumeration | status not stated | [19:19](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1159s) |  |
| Agent factory interface | status not stated | [20:17](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1217s) |  |
| Agent metadata interface for runtime configuration | status not stated | [22:32](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1352s) |  |
| Take agent task execution interface | status not stated | [23:18](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1398s) |  |
| Optional interface methods | generally available (roadmap [Evolve AL interfaces with default implementations](../features/573352.md)) | [23:50](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1430s) | "This is a feature that's coming up. Actually, Quintin here worked on it, and he's a he's ready to release it almost" ([23:50](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1430s)) |
| Complex agent scenarios with user intervention | status not stated | [24:33](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1473s) |  |
| AI evaluation tools for agents | status not stated, demoed | [26:54](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1614s) |  |
| Copilot credit consumption visibility in tests | status not stated, demoed | [27:29](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1649s) |  |
| Agent instance experimentation | status not stated, demoed | [27:42](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1662s) |  |
| AI evals suite page | status not stated, demoed | [28:07](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1687s) |  |
| Budget controls for agent tests | status not stated, demoed | [30:14](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1814s) |  |
| Agent task drill-down and troubleshooting | status not stated, demoed | [31:22](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1882s) |  |
| Copilot credit consumption visibility for non-admins | status not stated, demoed | [33:33](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2013s) |  |
| Test run history and trend tracking | status not stated, demoed | [34:03](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2043s) |  |
| Turn-based test design pattern | status not stated, demoed | [34:27](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2067s) |  |
| AI development toolkit evaluation extension | status not stated, demoed | [36:19](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2179s) |  |
| Agent test library | status not stated, demoed | [36:42](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2202s) |  |
| Data-driven agent test approach | status not stated, demoed | [35:51](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2151s) |  |
| AI test suite XML configuration standard | status not stated, demoed | [38:36](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2316s) |  |
| AI test type code unit | status not stated, demoed | [39:58](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2398s) |  |
| AI test type property | status not stated, demoed | [39:16](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2356s) |  |
| Agent test context code unit | status not stated, demoed | [40:42](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2442s) |  |
| YAML-based test data sets | status not stated, demoed | [41:40](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2500s) |  |
| Run turn and wait method | status not stated, demoed | [42:59](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2579s) |  |
| Data-driven agent test framework | status not stated, demoed | [44:01](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2641s) |  |
| Agent debugging with break on next | status not stated | [44:59](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2699s) |  |

A status with a roadmap link comes from the Microsoft 365 roadmap feature this part of the video covers (matched by Haiku; links Opus dropped are not used); other statuses need a status word in the video itself.

## AL objects mentioned

As heard in the captions. A name that matches one object page by exact type and name links to it; the others stay as named.

- other "agent" at [1:51](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=111s)
- other "agent models list" at [6:22](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=382s)
- other "task template" at [8:07](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=487s)
- other "task log" at [8:53](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=533s)
- other "agents list" at [13:23](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=803s)
- enum "agent metadata provider" at [19:35](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1175s)
- interface "factory" at [20:17](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1217s)
- interface "agent metadata" at [22:32](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1352s)
- interface "take agent task execution" at [23:18](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1398s)
- page "AI evals suite pages" at [28:07](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1687s)
- codeunit "agent test context" at [40:42](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2442s)
- codeunit "AI test context" at [41:23](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2483s)
- [codeunit 4321 "Agent"](../objects/codeunit/4321.md) at [43:22](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2602s)
- [page 4316 "Agent List"](../objects/page/4316.md) at [45:25](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=2725s)

Not found in BC28-30: enum "agent metadata provider", interface "factory", interface "agent metadata", interface "take agent task execution", page "AI evals suite pages", codeunit "agent test context", codeunit "AI test context".

## Quotes

- [0:04](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=4s) "Agents are no longer just experiments. They're becoming a first-party product inside of Business Central."
- [1:30](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=90s) "This is all available for you today. You can look at it in the client. Um from 27.4 and newer."
- [2:37](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=157s) "Maybe this instance of my agent needs to work in a certain locale. Maybe it's a US decimal points or it's a European decimal"
- [4:16](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=256s) "As you mature the agent, you need to come back and figure out if the agent is seeing too much or not seeing enough"
- [6:22](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=382s) "you wouldn't even see this agent models list. That would be for 28 where you have now this drop down"

## Disclaimers in the video

- [6:22](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=382s) subject-to-change: I expect that when we validate more and more models, we will increase the availability here
- [23:50](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1430s) coming-later: This is a feature that's coming up. Actually, Quintin here worked on it, and he's a he's ready to release ...
- [27:53](https://www.youtube.com/watch?v=MNwTt06ZxwY&t=1673s) preview: this is available in sandbox only since this is a testing tool, but you'd be running this before going to production

Presenters (as heard): Blazhe, Stefan, Quintin.