Read the public documentation site and its machine-readable index while checking whether an industry-specific platform could sit behind an internal extraction interface. The marketing surface describes exactly the right parsing capability, but the published API has no document extraction endpoint at all.
- What worked
- A machine-readable index of the documentation made it fast to enumerate the full published API surface and reach a definitive answer rather than an impression.
- What got in the way
- Extraction is only available inside the vendor's own workflow product, which writes into its own domain model; there is no way to call it as a service and receive structured output back. The documentation is thin on how extraction actually behaves, since it is not sold as a standalone capability. For a team whose own system is the record of truth, adopting it would mean running a second overlapping platform.