How to Add an AI Player That Reads Live Game State with Convai (Part 1) | Unreal Engine Tutorial

Hello everyone,

Ever wanted to build a game state-aware AI companion that can work alongside human players to solve complex puzzles? We are excited to release Part 1 of our two-part tutorial series on how to integrate a Convai-powered AI teammate inside Unreal Engine using Epic Games’ free Stack-O-Bot project!

In this tutorial, the orange bot represents the human player, while the pink bot is a Convai-powered AI teammate. Instead of following scripted dialogue trees, the AI teammate listens to user voice commands, understands tagged environment objects, receives live game state, and takes dynamic actions like pressing pressure plates, stepping on moving platforms, and assisting with puzzle logic.

:sparkles: What We Cover in Part 1

This is a step-by-step teaching tutorial where we build the AI companion workflow from scratch:

  • :robot: AI Teammate Blueprint Setup: Bring the AI player into the scene, set up the Convai Chatbot component, and enable player voice interactions.

  • :person_running: Navigation & Convai Actions: Configure pathfinding and movement actions so the AI teammate can follow the player and move to specific targets.

  • :high_voltage: Dynamic Context & Live Game State: Stream real-time gameplay updates to the AI’s Dynamic Context—allowing the character to know if a pressure plate is pressed or if a platform is moving.

  • :eye: Environment Object Tagging: Tag objects and interactive elements in the scene so the AI understands spatial context and object state changes.

  • :puzzle_piece: Puzzle Coordination: Test real-time human-AI collaboration to solve two-player puzzle logic.

(Note: In Part 2, we will extend this project with custom crate pickup and drop actions!)

:link: Project Files & Helpful Resources

We’ve provided project files, Delta files, and complete setup guides so you can follow along directly:

Prerequisite & Workflow Videos:

We’d love to hear how you plan to use Dynamic Context and live game state in your co-op games or AI companion projects. Drop your thoughts or questions in the replies below!

Happy developing!

1 Like