Foundation Model Switchboard with Policy-Based Intelligent Routing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Integrating generative AI foundation models into enterprise platforms is challenging due to the diverse range of models available, varying strengths and weaknesses, lack of standardization, and the need for context-specific optimization, leading to inefficient resource use and significant experimentation.

Innovation Solution

A switchboard platform that provides intelligent access, routing, and orchestration of foundation models based on policy parameters, enabling context-aware selection and management of models to optimize resource use and improve efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If enterprises integrate multiple diverse foundation models into their platforms, then the capability and versatility of the system is improved, but the device complexity and difficulty of management increase significantly

Engineering Contradiction:
Improvemodel capabilityVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent introduces a switchboard platform as an intermediary layer between applications and foundation models. This switchboard manages model routing, selection, and orchestration, shielding applications from the complexity of having multiple diverse models. The switchboard acts as a mediator that handles model-specific configurations and routing logic centrally.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The switchboard platform provides universal access to multiple foundation models through a single interface. It implements model-agnostic routing capabilities that can handle different model types (LLMs, vision models, audio models) uniformly, allowing one system to serve multiple functions and model families without requiring separate integration logic for each.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Reliability

If enterprises deploy multiple foundation models with different strengths and weaknesses, then the performance for specific tasks can be optimized, but the resource consumption and cost increase

Engineering Contradiction:
Improvetask performanceVSAvoidresource consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent implements model-specific routing policies that match particular foundation models to specific task types based on their strengths. For example, certain models are designated for text generation tasks while others are optimized for vision or audio tasks. This ensures that each task uses the most appropriate model for its specific requirements rather than a generic approach.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The switchboard dynamically adjusts routing decisions based on task parameters and model capabilities. It evaluates task characteristics (modality, complexity, domain) and changes routing parameters in real-time to select the optimal model, thereby optimizing resource utilization and cost efficiency while maintaining high performance for each task type.

Inventive Principle:
Principle #35Parameter changes

3Productivity

If enterprises implement context-aware model selection, then the efficiency and resource utilization improve, but the system complexity and difficulty of operation increase

Engineering Contradiction:
Improveresource efficiencyVSAvoidoperational simplicity
Core Design Contradiction:
ProductivityVSEase of operation

Solution Approach 1:

The switchboard implements automated model selection and routing that operates autonomously based on task characteristics. The system self-evaluates incoming tasks against model capabilities and routing policies, making selection decisions without manual intervention. This automation maintains high resource efficiency while simplifying operations for end users who don't need to understand the underlying model selection logic.

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP4647941A1Switchboard platform for foundation models
Publication Date: 2025.11.12 ACCENTURE GLOBAL SOLUTIONS LTD
  • EP4647941A1 patent drawingFigure 1
  • EP4647941A1 patent drawingFigure 2
  • EP4647941A1 patent drawingFigure 3

AI summary

Implementations of the present disclosure provide receiving, by an intelligent router of the switchboard platform, a first request from an application, the first request including at least a portion of a prompt and a set of policy parameters, selecting, by the intelligent router, a foundation model of a sub-set of foundation models at least partially based on at least one policy parameter in the set of policy parameters, determining, from a model registry of the switchboard platform, connection data for the foundation model, transmitting, by the intelligent router and through a model connector of the switchboard platform, a second request for processing by the foundation model, the second request being transmitted using the connection data and including at least a portion of the prompt, receiving, by the intelligent router, a response from the foundation model, and transmitting the response to the application.