docRefinery vs Google Document AI

Processors on a platform, or one extraction product

Document AI is built from processors. Enterprise Document OCR digitises pages. Form Parser, Layout Parser and the Custom Extractor pull out structure and entities. Pretrained parsers cover invoices, expenses, bank statements, W-2s, pay slips, US driving licences and identity proofing. Custom classifiers and splitters sort and separate mixed files. Each Google Cloud project creates its own processor instances, and you wire them together and handle the results.

docRefinery is one product rather than a kit. Classification, splitting, extraction, confidence, business rules and review arrive together, configured by your organisation instead of assembled in code.

Where Google Document AI is the better choice

• Compliance breadth. Google lists Document AI as in scope for its ISO 27001, 27017 and 27018 certifications, its SOC 1, 2 and 3 reports and PCI DSS. Its security page says Document AI is HIPAA compliant and FedRAMP High compliant, and that Google never uses customer data to train its Document AI models. docRefinery's SOC 2 Type II audit is in progress with no report yet, and we make no data-residency commitments.
• Google Cloud controls. Customer-managed encryption keys and VPC Service Controls fit an existing Google Cloud security model.
• Custom extraction with few samples. Google recommends starting with its generative foundation model, which it says needs roughly 0 to 50 or more training documents. Template-based training for fixed forms needs about three.
• Cheap OCR at volume. Enterprise Document OCR is free for the first 1,000 pages, then $1.50 per 1,000 and $0.60 per 1,000 above five million.

What docRefinery does that Google Document AI does not, or does differently

docRefinery is what you would otherwise assemble from processors. These are the differences that matter once the kit is in production.

• 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, across every document type rather than one custom processor.
• One output format for every document. Every field returns text, alias, field_confidence and location, and every table returns rows, so one parser replaces per-processor handling.
• A review queue in the product. Google lists Document AI's Human-in-the-Loop feature as deprecated since 16 January 2024, and suggests flagging low Custom Extractor scores for manual review, which leaves the review itself to you. In docRefinery, values under a per-field threshold go to a queue, and every correction stays on the run.
• Learns from corrections. A reviewer's correction becomes a suggested fix, tested on that layout's own documents before anyone applies it.
• Layout memory. Each sender's layout is learned once and remembered, and a vendor's redesigned template is caught as a new layout. Our guides to vendor layout changes and invoice data extraction go further.
• Long files, segmented. Long, multi-page files are handled at scale and separated into the documents inside them, with no chain of classifier, splitter and parser to wire up.
• Custom types with no hosting fee. With Document AI, a custom processor is charged for hosting as well as prediction: $0.05 an hour per deployed processor version, which Google's own example puts at $438 a year. docRefinery trains, hosts and re-trains classifiers for types outside its catalogue at no charge, and you pay only for extraction volume.
• Governed fields. Owner, admin and member roles decide which fields may be extracted, and a field missing from the page comes back null.
• Your network, not a cloud project. Managed cloud, your private VPC, or fully air-gapped on your own servers on Enterprise.
• Business rules. Checks such as line items adding up to the total run on every document, with versions you can replay before a change goes live.

How the pricing differs

Google prices by processor. Form Parser and the Custom Extractor are $30 per 1,000 pages up to a million, then $20. Layout Parser is $10 per 1,000. Custom classifiers and splitters are $5. The invoice, expense and utility parsers charge $0.10 per document of up to ten pages, and the bank statement parser $0.75 per classified document. Savings plans lower these rates. docRefinery is free up to 100 pages a month, and paid plans are custom, quoted to your volume.