---
id: post/demiliani-com/12036
type: post
title: Why not start improving Business Central APIs definitions for supporting AI?
summary: Business Central APIs lack AI-ready metadata that LLMs need to reliably call functions. The post explains how exposing API descriptions, field types, and OpenAPI specifications would help AI agents automatically discover and use the right endpoints without explicit function calling.
tier: community
language: en
tags:
  - api definitions
  - function calling
  - semantic kernel
  - openapi specification
  - ai metadata
  - llm integration
  - odata filters
system: integration
review:
  state: reviewed
  by: opus
  at: "2026-10-08T02:00:19.047Z"
  flags: []
generated:
  at: "2026-10-08T01:26:53.141Z"
  pipeline: 0.2.0
  prompts:
    extract-post: 1
  input_hash: 48f4fa2f31860249199ed0f59a68d0c943de57dddd77a5e02464904508a1737d
evidence:
  - kind: blog
    url: https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/
    title: Why not start improving Business Central APIs definitions for supporting AI?
    date: "2025-07-08"
    commit: null
    t: null
    quote: null
  - kind: blog
    url: https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/
    title: Why not start improving Business Central APIs definitions for supporting AI?
    date: "2025-07-08"
    commit: null
    t: null
    quote: An LLM-ready API is an interface designed for intelligent agents to use reliably and programmatically without the need for human translation.
  - kind: blog
    url: https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/
    title: Why not start improving Business Central APIs definitions for supporting AI?
    date: "2025-07-08"
    commit: null
    t: null
    quote: AI agents require access to comprehensive API schema definitions, including endpoints, parameters, and capabilities, to function effectively.
  - kind: blog
    url: https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/
    title: Why not start improving Business Central APIs definitions for supporting AI?
    date: "2025-07-08"
    commit: null
    t: null
    quote: It's time to make APIs the cornerstone of your AI strategy also in Business Central.
links:
  learn: []
  objects: []
  features: []
  topics: []
  localizations: []
  videos: []
  posts: []
  guidelines: []
post_id: "12036"
source_id: demiliani-com
source_name: Stefano Demiliani
url: https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/
published_at: "2025-07-08T07:08:17.000Z"
author: Stefano Demiliani
full_text: false
words: 1779
quotes:
  - text: An LLM-ready API is an interface designed for intelligent agents to use reliably and programmatically without the need for human translation.
    why_it_matters: "Defines the goal: APIs must be naturally understandable to language models without requiring human explanation or workarounds"
  - text: AI agents require access to comprehensive API schema definitions, including endpoints, parameters, and capabilities, to function effectively.
    why_it_matters: Core requirement for AI solutions in Business Central that prevents hallucinated behaviors and incorrect API calls
  - text: It's time to make APIs the cornerstone of your AI strategy also in Business Central.
    why_it_matters: Captures the central thesis that API design should prioritize AI consumption as a fundamental business strategy
code_objects_mentioned:
  - page SD Sales Invoice API
  - table Page Metadata
  - table Page Table Field
  - table Page Info and Fields
systems:
  - integration
  - copilot
  - development
versions_mentioned: []
preview:
  embeddable: true
  frame_url: https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/embed/
  image: https://demiliani.com/wp-content/uploads/2025/07/apiforai_00.png
  image_alt: null
  image_w: 488
  image_h: 325
  site_name: Stefano Demiliani
  favicon: https://demiliani.com/wp-content/uploads/2022/09/sd_wordpress_logo.png?w=32
  probed_at: "2026-10-07T11:51:59.609Z"
---

# Why not start improving Business Central APIs definitions for supporting AI?

[Read the post](https://demiliani.com/2025/07/08/why-not-start-improving-business-central-apis-definitions-for-supporting-ai/) · Stefano Demiliani (Stefano Demiliani, MVP) · 2025-07-08 · 1779 words · tier community · reviewed (checked by Opus)

> Business Central APIs lack AI-ready metadata that LLMs need to reliably call functions. The post explains how exposing API descriptions, field types, and OpenAPI specifications would help AI agents automatically discover and use the right endpoints without explicit function calling.

## Key points

- Current Business Central APIs were not designed for AI and lack descriptions that LLMs need to understand field meanings and relationships
- API page Description properties should be exposed in Page Metadata table to pass dynamic API context to language models
- OpenAPI specification support should be offered out of the box for all APIs (standard, custom, third-party) to enable automatic schema discovery
- AI agents require structured metadata including endpoints, parameters, request/response formats, and field explanations to function effectively without hallucinating

## Quotes

- "An LLM-ready API is an interface designed for intelligent agents to use reliably and programmatically without the need for human translation." (Defines the goal: APIs must be naturally understandable to language models without requiring human explanation or workarounds)
- "AI agents require access to comprehensive API schema definitions, including endpoints, parameters, and capabilities, to function effectively." (Core requirement for AI solutions in Business Central that prevents hallucinated behaviors and incorrect API calls)
- "It's time to make APIs the cornerstone of your AI strategy also in Business Central." (Captures the central thesis that API design should prioritize AI consumption as a fundamental business strategy)

## AL objects mentioned

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

- page "SD Sales Invoice API"
- table "Page Metadata"
- table "Page Table Field"
- table "Page Info and Fields"

Not found in BC28-30: page "SD Sales Invoice API", table "Page Metadata", table "Page Table Field", table "Page Info and Fields".

## Context

- Features: API metadata exposure, OpenAPI specification support, Description property for AI, Dynamic API discovery, Schema validation, Field type documentation

Source: Stefano Demiliani, community blog. Summary, key points and quotes are derived (CONTENT-NOTICE.md); read the original for the full text.
