11 posts about ai document processing.
How modern document AI works and how to put it into production, from the parsing models to a pipeline with measured accuracy, a review queue, and a known cost. Each step builds on the one before it, using the same synthetic invoice where it can.
The vendor shortlist and the five questions to take to every document AI sales call.
What changed when parsers moved from OCR engines to vision-language models, and what that risks.
The four operations, what each returns, and which one your problem needs.
Install Marker, read its output, then make the same calls through the Datalab API.
How Extend ties schemas, evaluation sets, and review workflows together, with a worked invoice schema.
Five vendors compared on one invoice task from their docs and published prices.
Wire extraction, validation, and routing into a working invoice pipeline in n8n.
Score extraction field by field against ground truth before you trust it.
Decide what a person reviews, what they see, and how corrections flow back.
What the pipeline costs at volume once review time is counted, with the Lab calculator.
The terms used across the series, defined in one place.
Install Marker, convert a PDF, read the output, then do the same through the Datalab API. Every command comes from the official READMEs and docs.
Published · Updated
Which documents a person should check, how many to spot-check, how many reviewer hours you need, and how corrections make the system better.
Published · Updated
Per-page prices for Datalab, Extend, Azure, Google and LLM APIs as of September 2026, and a worked example showing why human review is most of the bill.
Published · Updated
What changed when document parsing moved from OCR to AI models that read the whole page, with dated examples from Datalab, Extend, Azure and Google.
Published · Updated
Thirty document AI terms, from OCR and parsing to precision, review thresholds and per-page pricing, each linked to the post where it matters.
Published · Updated
How Extend's extraction schemas, test sets and workflows fit together, from its docs, with a worked schema for one made-up invoice.
Published · Updated
Extend, Datalab, Google, Azure and Mistral compared on one disclosed invoice task, using only documented features and published prices.
Published · Updated
Check how accurate your invoice extraction really is, field by field, with a small answer key and a tested Python script that scores every field.
Published · Updated
The four document AI operations, what each returns, and when each is the right tool, walked through one synthetic two-page invoice.
Published · Updated
Import a tested n8n workflow that validates extracted invoice fields and routes exceptions to review, then swap in Gmail, Datalab and Sheets.
Published · Updated
Datalab, ABBYY, Rossum, Google and Azure for pulling data out of documents, and the questions that decide which one fits your team.
Published · Updated