Extract structured data from invoices, receipts, delivery notes, and finance documents — then send it where your systems need it.
typical extraction time per document
standard document types live
customer documents used to train shared models
From the moment a document arrives to the moment its rows are in your accounting system. No retyping in the middle.
Upload through the web, the API, or forward your supplier inbox to a DocReadi address. PDFs, images, scans, and most modern document formats. WhatsApp works too, once you connect your own Meta Business number — email forwarding needs no setup.
We extract every field on the document and check it against what we know about the counterparty. Anything that drifts — new bank account, new address, a total that doesn't match the line items — gets flagged for review.
Approved documents flow out as JSON via the API, webhooks on every state change, CSV or XLSX exports, or rows in your accounting system. You decide which.
Custom templates if your stack uses something else. The fields below are the ones we extract by default — the structured rows your accounting system actually consumes.
Line items, totals, VAT, supplier bank account, payment terms.
What your customers received from you. Useful for AR reconciliation.
Refunds and adjustments, mapped to the original invoice.
Line items and totals against expected supplier responses.
Quantities received against the line items on the related invoice.
Store, date, total, line items where present, VAT.
Every invoice teaches us a little more about the people you do business with. Drift detection comes free.
The first time we see a supplier, we record their bank details, address, and contact info. Every future invoice gets checked against that record. New bank account on invoice #17? You'll know.
We don't ask you to confirm the same thing twice. After three independent invoices show the same change, we promote it to "verified" automatically.
When a supplier's bank account or address changes, we flag the specific field — not the whole document. You see exactly what's new and decide whether to confirm or reject.
However you want to consume the structured rows — we have it. Mix and match across your tenant.
GET, POST, webhook events. Idempotent endpoints. Structured API reference. Bearer-token auth. Rate limits documented.
Read the API quickstart →We POST to your endpoint when a document is extracted, flagged, or approved. Retries with exponential backoff. Signed payloads.
CSV and XLSX in one click. Filter by counterparty, date range, status. Scheduled exports to S3 or email.
Your team reviews flagged documents inside DocReadi. Approve, reject, or edit. Changes sync back through the API.
Native today: Xero, via CSV export. Everything else: webhooks plus CSV / XLSX cover the gap — any system that can consume an import or a signed webhook will work. Custom field mappings live in your tenant settings.
What we do. Read business documents. Extract structured fields. Learn counterparties over time. Detect drift in bank accounts and addresses. Hand the structured rows to your stack.
What we don't. Approval workflows. AP automation. Reconciliation. Payment release. We feed the systems that already do those — Xero today, and any other system through webhooks and CSV / XLSX exports.
If your question isn't here, ask us — we keep this list short on purpose, and the answers stay current.
No. Your tenant's documents stay in your tenant. We do not use customer documents to train shared models. The "0" you see at the top of this page is real.
We tell you. The document is marked Failed with the specific reason — encrypted PDF, low-resolution scan, unsupported language. You re-upload, edit manually, or contact support if you think we got the failure wrong.
We host in the EU today; multi-region is on the roadmap as we grow. Tenant data lives in Postgres with row-level security between tenants — no shared schemas. Sensitive fields are encrypted at rest. Two-factor authentication is available on every account. Webhooks are HMAC-signed. POPIA and GDPR Article 22 — talk to us if you have a specific compliance question.
We report extraction confidence per field, not per document — there isn't a single accuracy number that's honest. By default every document lands in your review queue; the per-field confidence scores just tell you which to look at first. When you're ready, turn on auto-approve per document type and set the threshold that clears a document without a human — it's opt-in, not something we decide for you.
Native: Xero, via CSV export. Anything else: webhooks plus CSV / XLSX export cover the gap — most accounting systems and ERPs accept one of those today. Custom field mappings are configurable in tenant settings.
Yes. A structured API reference — endpoints, params, and response shapes — at docreadi.com/api/docs-guide, plus a machine-readable JSON version at /api/guide for tooling. Bearer-token auth, signed webhooks, idempotent POST endpoints. Free during evaluation.
Self-serve, per document processed. Free 90-day trial (50 credits) — no card required. Top up anytime; top-up credits stay valid for 12 months. See full pricing for Standard / Pro / Firm / Enterprise tiers.
DocReadi reads documents and gives you structured rows. AP automation tools handle approvals, payments, and ledger posting. We're the layer underneath — we feed AP automation tools (and your ERP) the data they need to do their jobs.
Free 90-day trial — 50 credits, no card.