Agents are the next big thing in AI. Some define these “agents” differently from others, but the general idea is, they’re AI-powered tools that can perform tasks autonomously.

The agent hype has reached a fever pitch, but one startup was relatively early to the game: LlamaIndex. Founded by former Uber research scientists, Jerry Liu and Simon Suo, in 2023, LlamaIndex allows developers to build custom agents over unstructured data.

“LlamaIndex started as a toy open-source project in November 2022,” Liu told TechCrunch. “I became deeply interested in understanding how large language models (LLMs) could be used on top of proprietary data outside their training set, and built an initial set of tools enabling developers to index and include data in their LLM apps.”

Using LlamaIndex’s open-source software, which has racked up millions of downloads on GitHub, developers can create custom agents that can extract information, generate reports and insights, and take specific actions. LlamaIndex provides data connectors and utilities like LlamaParse, which transforms unstructured data into a structured format that can be used for particular AI applications.

While there are other open-source frameworks to build AI agents out there, LlamaIndex is differentiated by its suite of data ingestion, data management, and data indexing and retrieval solutions, Liu said. It can connect data from files like PDFs and PowerPoint presentations, as well as apps such as Notion and Slack, with an agent.

Salesforce, KPMG, and Carlyle are among the companies using LlamaIndex today, Liu said.

“All of these competing solutions solve specific problems at different parts of the generative AI stack, but then it’s the developer’s responsibility to piece together fragmented solutions to create a working agent,” Liu added. “This is a significant pain point that hampers shipping agents to production. LlamaIndex made it our mission to deliver the most secure, accurate, and easy-to-use platform for building end-to-end knowledge agents.”

LlamaIndex’s next chapter is an enterprise service built on top of the company’s open-source offerings. Called LlamaCloud, it lets customers create cloud-hosted agents that can work with and manipulate unstructured data in a variety of formats.

LlamaCloud can be deployed via a software-as-a-service installation or in a virtual private cloud, and comes with features including role-based access control and single sign-on, Liu said.

In part to help fund LlamaCloud’s development, LlamaIndex recently raised $19 million in a Series A funding round that was led by Norwest Venture Partners, and saw participation from Greylock as well. The new cash brings LlamaIndex’s total funding raised to $27.5 million, and Liu says that it’ll be used for expanding LlamaIndex’s 20-person team, and product development.

“We have sufficient runway to take us through initial commercial expansion of our platform,” Liu said. “We’re betting on a future where developers play a big role in delivering GenAI applications within the enterprise.”