Structured output is forcing a language model to return data that conforms to a defined schema — typically JSON with specified fields and types — so the result can be parsed and used by software reliably instead of as free text.
Free-form text is hard for programs to consume. Structured output constrains the model to emit a valid object matching a schema (via tool calls, JSON mode, or response-format constraints), so downstream code can rely on the shape and types.
It is the bridge between an LLM and a deterministic system: the model reasons in language but hands back machine-readable data.
Instead of a sentence, the model returns `{"ticker":"NVDA","net_insider_usd":-4200000}` that an application can act on directly.
Arkolith tools return clean structured data, so an agent gets typed rows it can compute on — not prose it has to re-parse.
Arkolith turns this into live, sourced data your agent can query — SEC filings, insider activity, and market data behind one key, every datapoint traceable to its origin.