Skip to main content

ARC Agent Definition Language

Agent Definition Language (ADL) – Simplifying AI Agent Design


Why it matters

Traditional prompting for AI agents is complex, inconsistent, and brittle.

Each new LLM release or behavior change requires re-tuning prompts—a process that is time-consuming and hard to scale.

ADL (Agent Definition Language) solves this by introducing a structured, standardized, and model-agnostic way to define agent behavior.


Key Benefits

1. Business-Friendly

  • Anyone not just technical experts—can define and manage agent behavior.
  • Reduces reliance on prompt-engineering specialists and engineers.
  • Business requirements are defined directly by business in the Agent ( highlight this ) especially useful in enterprises where complex reqs and overlaps impact time to market and need predivatlbit

2. Standardized & Scalable

  • Encapsulates behavior into use cases with clear steps, solutions, and fallbacks.
  • Provides consistency across models and across customer scenarios.

3. Resilient to Model Changes

  • Decouples agent logic from raw prompts.
  • When models evolve, ADL ensures minimal rework.

4. Extensible & Flexible

  • Supports conditionals for personalization.
  • Enables tool calls (@function()) for dynamic execution.
  • Allows use case references to link complex workflows.

5. Operational Reliability

  • Built-in fallbacks prevent failure loops.
  • Structured validation of tools and references ensures robustness.

Business Impact

  • Faster time-to-market for new agent behaviors.
  • Lower cost of maintenance as agents evolve.
  • Greater adoption since business teams can contribute directly.
  • Future-proofing against LLM drift and vendor lock-in.

👉 Think of ADL as the “agent behavior standard”—the same way SQL standardized database queries, ADL standardizes how we define intelligent agents.