We’re building a new kind of data product for a new kind of investment process

LLM agents are increasingly being used to execute complex, market-wide investment strategies. These agents are expected to inspect, filter, join, rank, and verify across thousands of companies and years of filings in search of overlooked opportunities.

For these kinds of agentic tasks, the bottleneck is not the model or the agent harness, it's the data.

Traditional APIs and MCP servers are useful for known lookup tasks, but they are a poor interface for exploratory market-wide research. They require the agent to make repeated network calls to gradually assemble the local data before the real analysis can begin.

Agents do their best work when the data they need is already in the workspace. They can search it, reshape it, query it, write scripts against it, and iterate without waiting on a remote service.

So we're building an agent-first data service that gives agents the data they need, where they need it, in the format they need.

Here's how it works.

The Chadwin CLI installs a local data pack anywhere your agent runs: your laptop, a sandbox, a cloud VM, or a dedicated Linux machine. This is a versioned, structured dataset designed specifically for agent workflows.

For our beta launch, the data pack contains insider insider buying and selling for the entire US market. Over the next few months, we're going to rapidly expand the data offered, including 10+ years of normalized financial income statements, balance sheets and cash flow statements, including derived metrics, all available as local data.

The agent can work with this data using the tools it already knows how to use: rg, find, jq, awk, shell pipelines, DuckDB, Python, Polars, pandas, CSV readers, Parquet readers, and custom scripts.

When deeper evidence is needed, Chadwin APIs fetch the full source artifacts: 10-Ks, 10-Qs, 8-Ks, Form 4 filings, exhibits, transcripts, and other documents. The API is still important, but it's used for what APIs are good at: retrieving specific evidence, not coordinating open-ended search.

Behind the scenes, Chadwin runs a data pipeline that prioritizes agent ergonomics. We fetch, transform, normalize, and package financial data with attention to searchability, token efficiency, provenance, and local analytical performance.

We think your agents will love it.