Intelligent Document Processing (IDP) with AI
Intelligent document processing uses OCR and a language model to read invoices, forms, contracts and scanned PDFs, pull out the fields you need and send them to your systems, with a person checking low-confidence items.
In short
- Scope: one document type or family (for example supplier invoices, intake forms, leases) taken from inbox or folder to a filed, validated record.
- Price: $9,800 CAD + HST for the six-week AI Pilot Sprint. Model and OCR usage is billed to you by the vendor.
- Accuracy is measured on your own documents before go-live, field by field, and low-confidence fields go to a person.
- Handles scanned pages (handwriting is tested case by case), bilingual English/French documents, and outputs to SharePoint, accounting, CRM or case systems.
- Personal information in the documents is mapped before the build, with hosting chosen for PIPEDA and Quebec Law 25 obligations.
Your situation
Document work is a strong first AI project because the cost is easy to see: people retyping what is already written. Typical cases:
- Accounts payable by hand. Supplier invoices arrive by email and mail; someone keys vendor, amounts, GST/HST and GL codes.
- Intake packages. Applications, referrals or onboarding forms arrive as PDFs and scans, and staff build the file manually.
- Contract and lease review. Staff search long documents for dates, renewal terms, parties and obligations.
- Scanned archives. Years of paper and image-only PDFs that nobody can search.
What gets automated
A typical IDP workflow has five steps. The AI handles the reading; rules and people handle the decisions.
| Step | What happens | Who checks |
|---|---|---|
| Capture | Documents arrive from email, scanner, portal or folder | Automatic |
| Read | OCR plus a language model identifies the document type and reads it | Automatic |
| Extract | Fields are pulled into a structured record | Low-confidence fields go to staff |
| Validate | Values are checked against rules and your systems (vendor list, PO, totals) | Mismatches go to staff |
| File | Record is created in the target system, original filed in SharePoint or DMS | Sample audits |
For inbox-driven processes that go beyond documents, see AI workflow automation. When the target is an older system with no modern API, see LLM integration with legacy systems.
What the AI Pilot Sprint includes
One document workflow in production. One document type or family, end to end, into your real system.
An accuracy test on your documents. Field-level results on a sample, reported before go-live.
Review queue and thresholds. Staff see the document next to the extracted fields and confirm or correct.
Logging and a runbook. Every document, field and correction is logged. Corrections feed the next round of tuning.
Training for the owners. The people who handle the documents learn the review screen and the exception process.
How long it takes
Six weeks. In our experience it usually splits into one week for samples and access, three weeks of build and accuracy testing, and two weeks of live use. It depends on the number of layouts, scan quality and the target system's API.
What it costs
The AI Pilot Sprint is $9,800 CAD + HST for one document workflow. OCR and model usage are billed by the provider to you, so the per-page cost is visible from the first day.
To compare several document flows before choosing, start with the AI Opportunity Audit at $2,900. See the pricing page for all packages.
Why Remolda for document processing
- Measured on your documents. Accuracy is tested on your sample before go-live.
- People stay in control. Review thresholds are set from measured accuracy.
- Your systems, your storage. Workflows, logs and files stay in your tenant.
- Privacy in the design. Personal information is mapped under PIPEDA and Quebec Law 25 before any document is processed.
- Fixed price and scope. Six weeks, $9,800, one workflow.
How the work runs
The Sprint is the Implement step of the Remolda Cycle (Audit → Strategy → Implement → Empower → Evolve).
- Sample. You share 50–200 representative documents, including the difficult ones.
- Build. Extraction, validation rules and review screen.
- Accuracy test. Field-level results on the sample, thresholds agreed.
- Live use. Documents flow into the real system with review on.
- Review. Time per document and error rate against the baseline; go, adjust or stop.
Frequently asked questions
What is intelligent document processing?
Intelligent document processing (IDP) is software that reads documents: it recognizes the text (OCR), understands the layout and meaning with a language model, extracts the needed fields, checks them against rules and passes them to the next system.
How much does AI document processing cost?
Remolda's six-week AI Pilot Sprint costs $9,800 CAD + HST for one document workflow in production. Per-page model and OCR costs are paid directly to the provider and depend on volume and page complexity.
How accurate is AI data extraction?
Accuracy depends on the document: clean digital PDFs extract more reliably than faxed or handwritten pages. We measure field-level accuracy on a sample of your documents before launch and set review thresholds from that number.
Which documents are good candidates?
High-volume documents with a known set of fields: supplier invoices, receipts, application and intake forms, purchase orders, bills of lading, leases and standard contracts, insurance claims and identity documents.
Can it handle French and bilingual documents?
Yes. Current language models read English and French, and mixed-language documents are common in Canadian workflows. We test on your bilingual samples before go-live.
Where are the documents processed and stored?
In your environment where possible: your SharePoint or cloud storage, and model endpoints in Canadian regions when the provider offers them and your obligations call for it. The data flow is documented before the build.
How long does it take?
The Pilot Sprint runs six weeks. In our experience the first week is spent collecting a representative document sample and access; the timeline depends on sample quality, the number of layouts and the target system's API.
Sources
- Office of the Privacy Commissioner of Canada — PIPEDA in brief
- Microsoft Learn — Foundry models sold by Azure: region availability
Facts checked:
Related services
Approach phases
Related insights
LLM Integration into Existing CRM and ERP Software in Canada: Architecture, Data Residency and Cost
Mitacs AI Advantage: Ottawa Commits $162M to 10,000 AI Work Placements
AI for Canadian Municipalities: Where It Works in 2026
Talk to an AI transformation consultant
A 30-minute call: you describe the situation, we tell you what to do first and what it would cost.
Book a 30-min call30 minutes. English or French.