MemGPT, a toolkit for developing AI agents, has raised $10 million at a $70 million valuation, according to details seen by Business Insider.
The round was led by Felicis Ventures, according to four sources familiar with the matter, and one of the sources said it was the company's first funding round.
MemGPT helps users “build LLM agents with long-term memory and custom tooling,” according to its associated GitHub page. The MemGPT project has received over 11,000 stars on GitHub, indicating the project's popularity among developers.
MemGPT and Felicis Ventures did not respond to requests for comment.
According to one source, MemGPT is an agent tuned for short-term and long-term memory: short-term memory provides the LLM agent with immediate context for a single session, while long-term memory acts as a knowledge base for the agent to recall previous context.
The startup's website highlights MemGPT's key features, including adaptive and long-term memory, a wide context window, and unlimited data. Developers can also use MemGPT to build multiple agents for multiple users.
Sarah Woodards and Charles Parker, PhD graduates from UC Berkeley, founded the company and co-authored the research paper behind it, which explains that the capabilities of large language models are limited by their context window (the amount of information an LLM can process at once). For example, LLMs struggle to analyze large documents with hundreds of documents.
To address this issue, MemGPT developed an agent framework, or toolkit for building AI agents, that focuses on memory management, allowing the AI agents to process much more information than usual. The framework uses an approach inspired by “traditional operating systems,” enabling LLMs to analyze large documents and remember long conversations, even when the information exceeds the model's normal processing limits.
Emergence, which has raised about $100 million, is a multi-agent framework for automating knowledge work in the enterprise. Other startups, including Crew AI and Phidata, are also looking to accelerate multi-agent systems, an increasingly hot investment area.
Big tech companies are also getting into the space: Microsoft's AutoGen and Semantic Kernel are open-source toolkits for developing AI agents, and Amazon Bedrock allows developers to access and build on a variety of underlying models through a single API.
