This page aims to explain the main differences between Optimization Engine and classic Machine Learning modeling and how Optimization Engine interacts with other Pricefx modules.

Advantages

Optimization Engine, once fully configured, combines powerful, transparent segmentation and guidance with AI-based multi-constraint full-price waterfall optimization.

Machine Learning Segmentation and Predictive modeling	MAAI-based Optimization Engine Full Price Waterfall Optimization
Data-driven or rules-based / Predefined, managerial segmentation model Calculation of optimized deal guidance based on: Margin or discount based distribution (B2B) Price elasticity (B2C) Identification of Cross-Sell and Upsell opportunities ℹ Machine Learning is not part of the demo solution out-of-the-box.	Multi-Agent Artificial Intelligence (MAAI) powered engine to simultaneously optimize any price waterfall element at any level of granularity  Glass box and interpretable AI Considering multiple constraints and business strategies Understanding interactions and indicating which constraints are impacting the optimization results Simulating results

Background

Optimization Engine module is built around Optimization Engine backend which is a Multi-Agent System to optimize prices, discounts, or any continuous values for any user-defined criteria, such as margin targets, revenue targets, and custom business rules. Optimization Engine backend is a giant computation graph representing all the variables, computations, and criteria in the problem and their relationships. Each node of the computation graph is an autonomous agent. At runtime, agents cooperate to gradually converge towards the values that satisfy all the criteria, or towards the best compromise in case of conflicting criteria (such as increasing margin but also keeping prices as stable as possible).

Position within Pricefx Platform

The following diagram focuses on Optimization Engine position within our cloud-native 360° pricing platform:

Accelerator and Setup

The current POAI Accelerator is a key part of the Optimization Engine. It deploys a set of Logics to tackle a “Waterfall Optimization” problem as described in PO.AI Accelerator Overview.

To set up Optimization Engine, you need to first understand the customer’s domain and requirements, and turn them into a 'problem description' that is fed to the Optimization Engine backend. Each instance of Optimization Engine has its own specific problem description file which is the result of problem modeling. It can be configured based on POAI Accelerator, documented in POAI Accelerator: an accelerator for PO.AI module.