Skip to content

Documents read for you, you check the exceptions.

Invoices, contracts and claims read and keyed without a hand on them. 96% extraction accuracy, and a person reviews the rest.

Still re-keying invoices, contracts and applications by hand? Your team is the OCR, and that is the problem.

  • 96% accuracy, people review the rest
  • Scans, PDFs and handwriting
  • Every field traceable to its page
Why it's been hard

Every document still needs a human. Even the ones that don't.

Invoices, contracts, applications, claims - 30-40% of admin time goes to someone reading a document and typing what it says into another system. It’s slow, error-prone, and the worst use of your most capable people’s time. RIVER builds the layer that reads it first.

Your process today

With RIVER

An invoice arrives as a PDF, and someone re-keys every line into the GL.

Line items extracted and mapped to your GL codes automatically.

A contract gets read once, then its key dates live in someone's memory or a spreadsheet.

Clauses, dates and obligations extracted and matched to your risk categories.

A claim or application is read cover to cover before anyone can act on it.

Fields extracted and confidence-scored the moment it lands, for review on the low-confidence ones.

High-confidence fields flow straight through; nothing enters your systems unreviewed on the rest.

Month-end means days of manual entry across hundreds of documents.

Hundreds of documents processed in minutes, in a batch your team starts and walks away from.

Nobody can say which document a figure came from, or who touched it last.

Every extraction, override and export logged and traceable to the source document.

Your team keeps the judgement call. The platform takes the reading and the re-keying.

What it does

What AI actually does here. Grounded in the work, not the hype.

Documents read, fields extracted, exceptions flagged - the platform does the reading and the re-keying, and leaves the call on anything uncertain to your team.

Read it, and map it to your systems

The two things every document needs first: getting the data out, and putting it where your systems expect it.

96%
extraction accuracy
Your platformExtraction
Invoice · INV-30142 · Northgate SuppliesReady · review
Invoice total$4,280.00
Due date14 Sep 2026
GL code match6120 · Freight
Line items12 extracted

One line item below the confidence threshold - flagged for a quick check.

ApproveCheck lineReject

Then trust the exceptions, and the scale

On that foundation, confidence scoring means your team reviews the fields worth reviewing, and the volume stops being the bottleneck.

hundreds
of documents processed per batch, minutes not days
Your platformConfidence

Claim CL-9031 · field confidence

Claimant detailsHigh · 98%
Incident dateHigh · 95%
Damage estimateLow · 61%
Policy numberHigh · 99%
Why it works

You don't start from scratch. We tailor what's proven to your documents.

RIVER is over 100 prebuilt modules, tailored to the documents you actually process in weeks - built to sit alongside the systems you already run, not replace them.

96%
extraction accuracy
30-40%
of admin time typically spent on manual entry
100%
audit-trailed, every field

Everything a procurement or compliance review asks for is in our Trust Centre

  • Your team reviews exceptions, not every document - high-confidence fields flow straight through.
  • Every extraction, override and export logged and traceable to the source document.
  • NZ-hosted, and your documents never train a model.
RIVER helped us validate our market and launch our platform, and they haven't stopped since. They've been with us every step.
Mike CollinsFounder, iMatch Sport · 2025
Questions

The questions people ask when they’re ready to move.

Ready when you are

Easy to start. Fast to prove. Built to scale.

Tell us about your documents, and we’ll take it from there. Prefer to look first? Ask Moana, or take the free Readiness Check.

Client stories

Teams like yours, in their own words.

A decade of work, across health, sport, finance and the public good.