Connect your data to AI agents
HQ: New York, NY, United States | Founded: 2022 | Employees: ~50–100 (est.) | Stage: Series A — $19M led by Norwest (March 2025); $27.5M raised total | Website: https://www.llamaindex.ai
LlamaIndex provides the data layer for AI agents. Its open-source framework connects LLMs to private data sources — documents, databases, APIs — handling ingestion, indexing and retrieval so developers can build question-answering and agent applications over their company's own knowledge. Tens of thousands of projects use it as the standard way to make unstructured data usable by models.
The company around the framework was founded by Jerry Liu and Simon Suo in 2022 and incorporated in 2023. Its commercial evolution reflects what customers actually struggled with: not retrieval algorithms but document chaos. LlamaParse, now generally available inside LlamaCloud, applies agentic AI to parsing — reading complex PDFs, tables, charts and slides the way a person would — and feeds clean structured content into agent workflows.
In March 2025 the company announced both milestones together: LlamaParse GA and a $19M Series A led by Norwest Venture Partners, bringing total funding to $27.5M. The pitch to investors was 'knowledge agents' — the layer that lets enterprises build agents on their own data with security and accuracy they can defend in procurement.
Every company wants AI that knows its own documents, and almost every company discovers the same problem: the documents are a mess. PDFs, scans, slide decks, spreadsheets and wikis hold the knowledge, but models can't use them until something parses, structures and indexes the content. LlamaIndex is that something — first as an open-source framework, now as managed infrastructure.
Jerry Liu (CEO) and Simon Suo (CTO) created the project in 2022 when 'GPT indexing' was a weekend hack, and it became one of the two default frameworks for building LLM applications over private data. The strategic move came in 2024–2025: rather than stay a library, the company productized the hardest part — document parsing — as LlamaCloud and LlamaParse, its agentic OCR and document-intelligence service. Wikipedia now describes the company precisely: agentic OCR and document intelligence infrastructure.
The $19M Series A led by Norwest (with Greylock, Databricks Ventures and KPMG participating) in March 2025 took total funding to $27.5M — modest capital against the size of the problem, deliberately.
The competition is the story's tension. LangChain owns more framework mindshare; hyperscalers sell their own retrieval tooling; and model vendors keep absorbing features. LlamaIndex's wager is that document understanding is a durable, unglamorous bottleneck — and whoever turns enterprise mess into clean, agent-ready context owns a tollbooth on every knowledge agent built after it.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki