Azure Cosmos DB - Globally Distributed Multi-Model Database
Azure Cosmos DB is a fully managed, globally distributed NoSQL and relational database service designed for low-latency, high-throughput …
Multi-Model
Compound AI System
An AI architecture that combines multiple models, retrievers, tools, and programmatic logic to solve tasks that exceed the capabilities of …
Compound AI Systems - Architecture Framework for Multi-Model Coordination
How compound AI systems combine multiple models, retrievers, tools, and control logic to achieve capabilities beyond what single models can …
Fallback Chain Pattern
Cascading model fallback strategy where failures or low-confidence responses trigger automatic failover to alternative models, ensuring …
Model Ensemble Patterns for AI Applications
Combining multiple models for improved accuracy, reliability, and coverage. Voting, cascading, and specialization ensemble strategies.
Model Tier Routing - Matching Request Complexity to Model Cost
Route AI requests to different model tiers based on complexity, cost sensitivity, and quality requirements. Reduce spend without sacrificing …
Multi-Model Routing Patterns
Strategies for routing requests to different AI models based on task complexity, cost constraints, and latency requirements. Router design, …
Orchestrator-Worker Pattern
An orchestrator LLM decomposes complex tasks and delegates subtasks to specialized worker models or agents, coordinating results into a …
Plan-and-Execute Pattern - Separating Planning from Execution in AI Agents
A two-phase agent pattern where a capable planner model creates a step-by-step plan, then delegates each step to cheaper, faster executor …