GoodAI develops Large Language Model (LLM) agents equipped with Long-Term Memory (LTM) systems, enabling these agents to learn continuously from interactions and environmental changes. This technology enhances cognitive automation in various fields, including gaming and business, by allowing AI to retain and utilize past experiences for improved performance.
Funding
Funding not disclosed
Founders
Product
Problem
Large Language Model (LLM) agents often struggle to retain information and learn from past interactions, limiting their ability to perform complex tasks that require long-term memory and adaptation to changing environments. Existing LLMs have limited context windows, which prevents them from effectively utilizing past knowledge to improve future performance.
Solution
GoodAI develops LLM agents equipped with Long-Term Memory (LTM) systems that enable continual learning from interactions and environmental dynamics. These agents store and integrate user messages, assistant responses, and environmental feedback into LTM for future retrieval, allowing them to learn instructions, skills, and user preferences. By expanding the context window of LLMs and making it dynamic, GoodAI's LTM system helps the LLM reason about its past knowledge and better integrate information in its conversation history. This approach enables the creation of AI agents capable of lifelong learning and improved performance in various applications.
Target Audience
The primary audience includes AI researchers, game developers, and businesses seeking to enhance AI agent capabilities through long-term memory and continual learning.
Features
- LTM system that expands the context window of LLMs, enabling continual learning.
- Integration of episodic, procedural, emotional, and relational memory for richer NPC behaviors.
- Ability to store and retrieve conversations, thoughts, plans, actions, observations, skills, and behaviors in a vector database.
- Context-aware responses through a combination of Long-Term Memory (LTM), Short-Term Memory (STM), and episodic memory.
- Dynamic memory processing that considers context, recency, importance, and relevance for optimal retrieval.
- Integration with OpenAI GPT-4 model, with the flexibility to switch to other LLMs, including local models.