Simple Agent Designer Documentation
Index

Overview

Simple Agent Designer is a Microsoft Dynamics 365 Business Central extension that lets you design, configure, and run AI agents to automate daily business tasks directly within your Business Central environment.

What is Simple Agent Designer?

Simple Agent Designer bridges the gap between modern AI capabilities and your Business Central data. Instead of manually processing invoices, categorizing journal entries, analyzing customer data, or generating reports, you can create AI agents that perform these tasks for you — using the same data and business logic already in your system.

The extension is published by Hougaard and integrates with 12 leading AI providers — including OpenAI (ChatGPT), Anthropic (Claude), Azure OpenAI, Google Gemini, xAI Grok, Mistral, and more — plus LM Studio for local model hosting. See Key Features for the full list.

How It Works

At its core, Simple Agent Designer follows a straightforward model:

  1. You design agents — AI-powered workers with specific roles, instructions, and capabilities
  2. You define workloads — tasks that describe what needs to be done, with optional data attachments
  3. Agents process workloads — using AI to understand, analyze, and act on the task
  4. Results appear in Business Central — journal entries, document updates, reports, and more

The Agent-Workload Model

Think of it like delegating work to a team of specialized assistants:

  • A Workload is a task description — "Process these invoices" or "Categorize these bank transactions"
  • A Workload Type is a reusable template — defining which agent handles the task, what data to use, and how to process results
  • An Agent (also called a Worker) is an AI-powered specialist — configured with specific instructions, an AI provider, and optional tools
  • The Orchestration Engine coordinates everything — routing tasks between agents, managing tool calls, and tracking progress

Built for Business Central

Simple Agent Designer is not a generic AI tool bolted onto Business Central. It is purpose-built for the platform:

  • Agents can create general journal entries, sales documents, and purchase documents
  • Agents can query and update Business Central tables directly
  • Agents can run reports and retrieve results
  • The tool system gives agents controlled access to your business data
  • All work is tracked and auditable through the workload system

Who Should Use This Extension?

Simple Agent Designer is designed for Business Central users and administrators who want to:

  • Automate repetitive data processing — such as categorizing transactions, matching bank statements, or processing incoming invoices
  • Leverage AI for business analysis — have agents analyze customer data, vendor performance, or financial trends
  • Create document workflows — automatically generate sales orders, purchase orders, or journal entries from structured or unstructured data
  • Process Excel uploads — split and process spreadsheet data row by row with AI assistance
  • Build custom AI workflows — chain multiple specialized agents together for complex multi-step processes

Key Concepts at a Glance

Concept Description
Agent (Worker) An AI-powered specialist with a defined role, instructions, and AI provider
Workload A specific task to be processed, including description and optional data
Workload Type A reusable template defining how a category of tasks should be processed
Tool A function that agents can call to interact with Business Central data
Orchestration The engine that coordinates agent execution and worker cooperation
AI Provider The AI service (one of 12 supported, such as OpenAI, Anthropic, Azure, Google Gemini, or LM Studio) that powers an agent

Architecture Overview

Simple Agent Designer uses an interface-based worker pattern for flexible, extensible agent execution:

  • Multiple worker types handle different specializations (general AI, data queries, journal entries, sales documents, purchase documents, AL code execution, and data modification)
  • A singleton orchestration engine manages the execution flow, worker chains, and task queues within a session
  • Integration events allow the tool system to be extended with custom tools
  • Background processing supports asynchronous workload execution

This architecture means you can start simple — with a single agent processing one type of task — and scale up to sophisticated multi-agent workflows as your needs grow.