AI agents for document extraction

Pull structured fields out of invoices, contracts, CVs and forms.

17 agents, each with its inputs, outputs and review level checked by a person.

Document extraction agents pull structured fields out of invoices, contracts, CVs, forms and reports, keep the location of each fact in the source, and mark fields that are unreadable or ambiguous instead of guessing. They return a structured record that a person or a system consumes. The listings record the document types each agent handles, how a schema is defined, what it returns and whether a person spot-checks results against the originals.

Choose by the documents and the volume. Invoices and forms suit template-driven agents; contracts and reports suit agents that read free text. In both cases the acceptance section matters: an agent that guesses a missing field is worse than one that flags it.

The agents

Groups that share agents with this one.

Questions

Can I define my own fields?

Most agents take a schema or a set of fields you define; the inputs section of each listing says how.

How accurate is extraction?

Vendors publish their own figures and the catalogue does not verify them. Prefer agents whose contract keeps unreadable fields marked rather than filled.