What we learned from our first 30 early-access teams
Six months into our early-access program, here's what surprised us about how back-office teams actually use document extraction - and what we built differently because of it.
Read articleThe Parseloom Blog
Document extraction, in practice
From the Parseloom team - what we learn building and running document AI for back-office teams.
Featured
All articles
How we measure extraction accuracy - and what 94% actually means
Accuracy numbers for document extraction are easy to game. Here's the methodology behind our internal benchmarks.
Why we built Parseloom API-first
Most document extraction tools are built around a GUI workflow. We decided the opposite - API first, UI second.
Getting your ERP team to actually connect a new tool
The extraction might work perfectly - but if your ERP admin won't open a port or add an integration, you're stuck.
The real cost of manual invoice keying - beyond just the hours
AP teams know manual data entry is slow. But the cost isn't just staff hours.
Building a custom extraction template for a document type nobody standardized
Purchase orders, remittance advice, customs declarations - here's how to build a Parseloom template for a document type we've never seen before.
Three patterns for getting extracted data into your ERP
Direct connector, webhook-to-middleware, or batch CSV drop - the right ERP integration pattern depends on your ERP version, IT team, and document volume.
Why structured field extraction is different from OCR - and when it matters
OCR turns pixels into characters. Field extraction turns a document into a structured record.
Why OCR keeps failing on bills of lading
Bills of lading defeat generic OCR for specific reasons: table structures with non-standard column widths, carrier-specific field placement, and handwritten annotations.
Eliminating the month-end invoice keying sprint
Most AP teams have a version of the same story: invoices pile up, month-end hits, everyone keys for a week straight.