About FactIQ
More and more research now runs through AI agents — screening companies, testing a thesis, writing a memo. But an agent is only as good as the data it can reach, and the open web hands it stale figures, paywalled sources, and numbers with no provenance.
FactIQ is the data layer that fixes this. It's a plugin for AI agents to access accurate macro data, market prices and company financials in one place. Your agent queries the data directly, computes in its own context, and publishes what it finds as shareable charts and reports — with every number traced to its source.
Our Story
We started FactIQ after running Defog.ai for three years, where we built the world's most popular open-source language models for data analysis and shipped AI-powered data pipelines with Fortune 500 companies including Toyota, Genmab, and AllianceBernstein.
Defog taught us where AI-powered analysis actually breaks — not in the reasoning, but in the data underneath it. Data in the wild is messy and rarely standardized: units differ, and definitions and measuring techniques change over time. A model doesn't notice any of that on its own — it will confidently compare two series that no longer measure the same thing.
We learned how to make databases legible to an LLM. This means reconciling units and frequencies across sources, tracking definition changes and revisions over time, and keeping the metadata a model needs — units, methodology, coverage — accurate and current.
That work became FactIQ. Over the last year, it's become our daily driver for investment research. Tell us what you think at founders@factiq.com.
The Team
Before Defog, we spent a decade working with data, policy and the institutions behind them: Manas worked on health and climate data at Reuters and The Washington Post; Rishabh led data initiatives at the National University of Singapore and Asian media conglomerates like Times Internet; and Medha worked on public policy and communications with companies including Microsoft, Huawei, and Grab.