Andrew Ng Releases New Short Course on Building Agentic Memory

DeepLearning.AI, a global edtech company, has released a free course titled “LLMs as Operating Systems: Agent Memory” to help people build agents. The company has partnered with Letta, an AI platform focusing on memory management for AI agents for the initiative. 

The course uses MemGPT’s research paper “Towards LLMs as Operating Systems” as a foundation. The research helps LLMs achieve a balance between storing a longer context in the input, and its processing speeds, through an agent. 

The course helps learners build an agent that can successfully manage the memory and context of an LLM using Letta’s open-source framework. 

These agents can summarise, persist, and retrieve conversation data efficiently, ensuring relevant information remains active whenever necessary and less important data is archived. 

It also teaches how to interact with a MemGPT agent and collaborate with multiple agents by exchanging messages and using shared memory blocks. 

“When I read the original MemGPT paper, I thought it was an innovative and important technique for handling memory for LLMs. The open-source Letta framework makes these ideas easy for you to use,” said Andrew Ng in a preview video for the course. 

In October, DeepLearning.AI released a course called “Multimodal RAG: Chat with Videos” in collaboration with Intel. As of today, DeepLearning.AI has released over 50+ short, and free courses on its platform, helping learners explore some of the best techniques for developing, and deploying generative AI products. 

Moreover, DeepLearning.AI also offers dedicated courses, and specialisations on broader topics like data engineering, machine learning, deep learning and so on. Over the years, Andrew Ng has been a dominant force in the democratisation of AI education, with over 7 million students using the platform to future-proof their skills. Earlier this year, Andrew Ng was also appointed to Amazon’s board of directors.