Overview

Utility Intelligence is a robust and powerful utility-based AI framework. Unlike traditional AI techniques such as BTs, FSMs, and GOAP, which rely on conditions for decision-making, our agents evaluate all possible decisions and choose the one with the highest score to execute.

Additionally, each decision is scored per target, ensuring that the best target is selected for the chosen decision. This approach makes our agents behave more naturally than those using other AI techniques 🤩.

Moreover, Utility Intelligence provides the Decision Making Preview feature, which allows you to preview which decision is selected directly in the editor without having to enter Play Mode. This feature can save you a lot of time when designing your agents’ behaviors.

How Utility Intelligence works¶

Every technique has its pros and cons. FSMs are great at handling transitions between states, but can only run one task at a time. Behavior trees can run multiple tasks sequentially or simultaneously, but they’re not strong at decision-making. Utility AI excels at decision-making but isn’t designed for executing tasks.

No single technique is good at everything. A well-designed framework usually combines multiple techniques, applying each technique where it works best to maximize benefits and minimize drawbacks. Therefore, in Utility Intelligence, we use:

• Utility AI for decision-making. It evaluates all decisions and selects the best decision with the highest score.
• Finite State Machines to manage and handle transitions from the current decision to the selected decision.
• Behavior Trees to execute the tasks of the selected decision, either in sequence or in parallel.