docRefinery vs Unstructured

Start with the output you need

Ask what happens to the data next. If a chatbot or agent will search it, you need clean, chunked, embedded text, and Unstructured's platform is designed around that: 20+ sources, 20+ destinations, chunking, embedding and loading on a schedule. If an accounts payable system will post it, you need a named value with a score and a trail back to the page. That is docRefinery. Its output feeds a business process, and a threshold you set decides what ships.

The line is not perfectly clean. Unstructured now has a structured data extractor that returns fields against a JSON schema, so it can do simple field extraction inside a pipeline. The gap lies in what surrounds those fields.

Where Unstructured is the better choice

For retrieval, we are not an alternative at all. We have no chunking, embedding or connector pipeline today, so if the requirement is a searchable knowledge base, choose Unstructured or something like it. Our retrieval pipeline, Parse, is planned, and webhooks and integrations are in design; the roadmap shows where they stand. It also leads in these areas:

• Breadth. Word, Excel, email, HTML and much more. Our list is shorter today; presentations and wider office formats are in development.
• Open source. The Python library is Apache-2.0 and its README says it will stay free, so engineers can start without a contract or an account.
• Compliance record. Unstructured lists SOC 2 Type 2, HIPAA, GDPR, ISO 27001 and FedRAMP High. That is a long list for any vendor in this category. docRefinery's SOC 2 Type II audit is in progress and there is no report yet; ask us for the latest status.
• Transparent pricing. A flat $0.015 a page after 10,000 free pages, with no sales call.

What docRefinery does that Unstructured does not, or does differently

docRefinery fits when a wrong value costs more than a slow one: a bank statement feeding a lending decision, or a passport in onboarding. A person needs to see which values to trust, and an auditor may ask where each came from.

• Business fields, not text. You get named values a system can post, where Unstructured's main output is elements, chunks and embeddings.
• You define the output. Each field has a plain-language instruction, a "Return as" name you choose as its output key, a type and a required flag, so values arrive under the names your system already uses.
• One output format for every document. Every field returns text, alias, field_confidence and location, and every table returns rows, so one parser covers invoices, statements and IDs.
• Line items as typed rows. Tables come back as typed rows, ready to post.
• Evidence on each value. A confidence score and a page and box come with each field by default, so an auditor can trace a figure to its source.
• Thresholds and review. A field that is absent comes back null. Anything below its threshold waits in a review queue, and corrections stay on the run and become suggested fixes.
• Governed fields. Your organisation, not each pipeline author, decides which fields may be extracted, with owner, admin and member roles.
• Layout memory. Each sender's layout is learned once and remembered, which matters little for a one-off archive and a great deal for monthly vendor documents. Our introduction to document data extraction covers how that differs from text preparation.
• Long files, segmented. Long, multi-page files are handled at scale and separated into the documents inside them, each classified into its type.
• Air-gapped on premises. Beyond private VPC, docRefinery runs fully air-gapped on your own servers on Enterprise.

Using both

The two can sit side by side. Unstructured prepares contracts and manuals for search, while docRefinery handles the invoices and statements that post to a ledger. If you only have one of those jobs, you only need one product.