AI Development
At Theil IT Management, we are exploring the frontier of agentic AI development — using teams of autonomous AI agents to design, plan, and build enterprise software on Microsoft Dynamics 365 Business Central.
The video below demonstrates a multi-agent system working near-autonomously to produce the first version of a Booking and Events Management application — from initial architecture through working functionality — with minimal human intervention.
The Agent Team
The system is composed of five specialized AI agents, each assigned a distinct role in the software development lifecycle:
| Agent | Role | Backed by |
|---|---|---|
| Cliff | Product Owner — defines requirements and priorities | Anthropic Claude |
| Frasier | Program Manager — coordinates the team and manages scope | Anthropic Claude |
| Sam | Junior Engineer — implements tasks and raises issues | Local Ollama model |
| Niles | Senior Engineer — reviews and refines implementations | Anthropic Claude |
| Daphne | Senior Engineer — leads technical delivery | OpenAI Codex |
Each agent communicates through a structured workflow, enabling the team to move through planning, implementation, and review cycles with a high degree of autonomy.
Development Sprints
The video covers the first two iterations of the project:
- Sprint 1 — Foundation: Establishing the core application structure, data model, and AL project scaffold in Business Central.
- Sprint 2 — Getting Started: Implementing initial seed data and onboarding functionality to make the application immediately usable.