Fabric.AI

Fabric.AI

Annie Chen and Anshul Poddar founded Fabric.AI in 2018 to bridge the decades-old gap between the volume of unstructured financial text produced daily and the quantitative models that need it as input. Both founders brought machine-learning expertise from prior roles in the technology sector, setting out to build a platform that does not simply search documents but extracts investable signals from the narrative structure of corporate disclosures, earnings calls, and sell-side reports. The firm operates from San Francisco. The platform processes millions of documents annually, covering earnings call transcripts, 10-Ks, 8-Ks, broker research, and regulatory filings. Its core output is a structured database of sentiment, topics, and forward-looking statements that quantitative hedge funds and asset managers pipe directly into their alpha models. Rather than offering a final trading signal, Fabric.AI provides the cleaned, time-stamped text layer that systematic investors need to avoid look-ahead bias and survivorship bias in backtesting. The firm sells data access, not investment advice, competing with legacy text-feed providers by offering deeper NLP-derived features such as management tone analysis and topic-trend tracking across thousands of companies globally. Coverage spans the US and major developed markets. Fabric.AI operates as a lean data and infrastructure company. It has raised venture funding, including a 2020 seed round from investors including Gradient Ventures, and has maintained a focused product roadmap centered on data quality, latency, and entity disambiguation rather than expanding into discretionary fund management. The team remains concentrated in engineering and data science roles. In May 2024, the firm continued to supply earnings transcript data and NLP analytics to quantitative hedge funds as part of its core commercial offering, reflecting a sustained posture as infrastructure for systematic equity strategies. Fabric.AI occupies an unusual space between financial data vendor and pure technology company. Unlike traditional data aggregators that employ armies of analysts to manually tag documents, the firm's entire value chain is automated NLP pipelines that scale with compute rather than headcount. This architecture lets it cover the full universe of US public companies with consistent metadata that systematic funds can trust for production trading models — a structure more akin to an AWS for financial text than a conventional research provider.