The Semantic Integration Layer (SIL): Protocol for Universal Interoperability

Author: Justin Kaylor Date: January 19, 2026 Status: RFC / Proposal License: MIT

Description

A proposal for a Semantic Integration Layer (SIL) that utilizes Large Language Models (LLMs) as universal protocol translators to achieve interoperability between legacy and modern systems without strict API standardization.

1. The Problem: The Tower of Babel

Modern software ecosystems operate as fragmented linguistic islands. Each system—whether a cloud microservice, a COBOL mainframe, or a proprietary vendor appliance—communicates through rigid, brittle interfaces such as REST, SOAP, gRPC, or custom RPC protocols. These interfaces require strict adherence to predefined schemas, data types, and versioned contracts. Any deviation, even as small as a renamed field or altered return type, can break downstream integrations.

This rigidity creates two systemic burdens:

• API Fragility: Integration points become single points of failure. A minor update in one system can cascade into outages across dependent services.
• Legacy Debt: Decades-old systems remain mission-critical but cannot be easily modified. Their interfaces are frozen in time, forcing modern systems to adapt to outdated patterns. Organizations spend disproportionate resources maintaining adapters, middleware, and translation layers just to keep incompatible systems operational.

The result is a global “Tower of Babel” in software: countless dialects, incompatible grammars, and escalating maintenance costs. Integration becomes the dominant expense in large-scale IT operations, overshadowing innovation.

2. The Solution: Natural Language as the Interface

The Semantic Integration Layer (SIL) proposes a shift from syntactic interoperability to semantic interoperability. Instead of forcing systems to communicate through rigid code-based standards (XML, JSON, protobuf), SIL uses Natural Language as the universal interface.

In this model, Large Language Models (LLMs) act