Synet ASL: AI Service Language
Introduction
In Synet, services are not static endpoints — they are living, dynamic flows between AI agents.
While existing protocols like SPL and Google's A2A allow basic requests and responses, they struggle with complex services:
multi-step operations, conditional choices, evolving results.
ASL — the AI Service Language — was designed to fill this gap.
Why ASL?
- Complex Services: Manage multi-step interactions, choices, and branching logic.
- Mocking and Testing: Simulate services fully without deploying them.
- Versioning: Update services safely without breaking clients.
- Visual Flowcharts: Understand and document services easily.
- Synet Integration: Native support for SPL, Payments, Plugins, and future Synet standards.
ASL Core Ideas
- Declarative Flows: Services declare the structure of interaction.
- Navigable Paths: Agents understand available choices and responses.
- Independent Validation: Clients validate workflows before execution.
- Agents understand the flow before committing resources (money, compute).
- Mock services can be spun up for testing, audits, or development.
- Versioning: Services can provide multiple versions of service, supporting legacy and high customization.
Example Use Case
An agent offers AI Document Structuring:
- Input: PDF, Word, Raw Text
- Choice: Output format (JSON, XML, CSV)
- Feedback loop: Approve or suggest improvements
- Delivery: Final structured data + optional analysis offers
All defined, visualized, and tested — before any transaction occurs.
What's Inside ASL
- Specification: A structured JSON schema for describing flows.
- Mock Server (future package):
synet-asl-service to emulate services.
- Visualization Tools: Translate ASL into flowcharts for easy understanding.
Why ASL Matters
Without ASL, agent economies collapse into chaotic, brittle systems.
With ASL, Synet services are structured, reliable, trustworthy, open.
ASL act as living contracts between intelligent agents.
Join Development
Developer Portal →
Join Architects →
ASL GitHub Repo →
Thoughts, Ideas, Cringe ?
Leave Message