# Uploader

> Uploader is a document verification platform. It takes in unstructured documents and data, splits, deduplicates and classifies them into clean records, extracts the data, runs your rules, cross-checks each file, flags exceptions to a reviewer and keeps a traceable evidence record that auditors, regulators and funders can review. Runs in the cloud or on your own infrastructure. Used for invoices and reconciliation, audit evidence, onboarding packs, supplier and contractor files, claims and regulatory submissions.

Key facts:
- Method, in order: Ingest, Clean & classify (split, deduplicate, classify, extract), Verify & comply (rules, registries, regulatory checks), Flag & record, Ask & act (agent across the whole dataset). The rules vary by context (a grant framework, a financial threshold, a regulatory requirement); the method is the same.
- Outcomes: each check returns Pass, Fail or Flagged for review, with the reasoning and source page.
- Relevant to any team that tests a high volume of documents against a checklist: finance and reconciliation, audit and compliance, HR and onboarding, contractor and supplier compliance, skills development, insurance, public programmes and immigration.
- 97.7% first-pass accuracy.
- Every page of every file is tested against your standard, not a sample. Uncertain results go to a person, and the auditor, regulator or funder makes the final decision. Uploader does not replace the auditor.
- Extracts names, ID numbers, dates and amounts into fields, including handwriting and poor scans, with a confidence score.
- Runs rules such as expiry dates, totals, VAT, thresholds and signatures, and cross-checks names, IDs and figures across documents.
- Records the source page, rule, reviewer and time for every result, and exports into your systems or each authority's template.
- An agent searches the whole database, with every answer linked to its source.
- Processes more than 1.5 million documents a month across 40+ organisations.
- Measured across teams running Uploader in production: 98% less time spent processing documents and 10× more documents with the same headcount.
- The Youth Employment Service (YES) onboarded 27,000 young people through Uploader in a single day.
- Most teams run their first live batch within four days; no IT department needed.
- Deployment: cloud by default. Where documents cannot leave your environment, the same platform runs on your own servers. On-premise deployments are licensed per installation.
- Sovereign AI security: choose where documents, models and results live (Uploader's encrypted cloud, a region you choose or your own servers). On premise, OCR, AI models and the verification engine run as containers on hardware you choose, with signed, versioned updates. Every AI result keeps its model output, confidence, rule and source page. POPIA and GDPR operator agreements; every view, edit and approval logged.
- Data cleanup service: send an archive as it is and receive one clean, verified dataset with the evidence behind every value.
- Uploader MCP: connect your own agent via MCP or API, including locally hosted models. A general-purpose agent can read a file; connected to Uploader, it can apply your rules to thousands of documents and return results with their evidence.
- Document audit: send any volume as it is and receive a record-by-record report within 72 hours.

## Pages

- [Automate complex document processes.](https://uploader.io/): Document AI for the manual work behind every process (one automated file journey: upload; split, deduplicate, organise and rename; extract and verify; compliance; reporting and insight), how Uploader ingests and cleans data in any format before applying rules, how it works in five stages (Ingest, Clean & classify, Verify & comply, Flag & record, Ask & act), dashboards, agentic consistent output at scale, solutions (finance and reconciliation, HR and recruitment, audit and compliance, CTOs and IT), use cases, the data cleanup service, the Uploader MCP, sovereign AI security (your documents, your models, your perimeter), pricing, FAQ and booking a 20-minute demo.
- [Data cleanup: messy data in, clean, verified records out](https://uploader.io/data-cleanup): send files, scans, exports and spreadsheets as they are; Uploader splits, deduplicates and classifies every file, extracts the data, verifies it against the source and returns a clean dataset, an organised file set, a gap list and an evidence trail linking each value to its document and page. Used for system migrations, audits, due diligence, year end, regulatory remediation and paper archives.
- [Supplier compliance and spend catch-up](https://uploader.io/backlog#supplier): year-end service for group finance and procurement teams. Ledger exports, AP extracts and supplier files taken as they are; duplicate suppliers merged, every spend line classified and each supplier checked against regional government sources (trading status, bank account, sanctions). You review every flag.
- [SETA, QCTO and YES evidence packs](https://uploader.io/backlog): for South African training providers, host employers and YES partners, corporate academies, SDF teams and B-BBEE consultancies. Send your documents as they are; Uploader sorts them, follows up on missing items, tests each pack against the programme requirements and prepares your 2026 submissions for approval. Audit report within 72 hours. Grant funding is lost in two ways: missing the deadline (WSP/ATR closes at midnight on 30 April 2027, with no grace period and no retrospective submission) or rejection (no proof of implementation, missing documents, expired accreditations or figures that do not reconcile). For priority bookings made before 5 February 2027, filing before the deadline is guaranteed provided the required documents can be obtained in time; acceptance remains the SETA's decision. Programmes: WSP/ATR, SETMIS, discretionary grants, learnership registrations, YES, Section 12H and B-BBEE skills development. Bookings close 5 February 2027.
- [From messy data to verified records, by industry](https://uploader.io/use-cases): finance, procurement and spend (multi-ERP ingestion, vendor master clean-up, spend classification and per-region supplier checks), audit and verification, HR and payroll, recruitment and placement, contractor and supplier compliance, skills development, supply chain, insurance and legal, public programmes, ESG and carbon reporting, maritime and customs, aerospace and manufacturing QA, clinical trials and pharma, immigration and visas, renewable energy development, media and royalties, and agriculture and food traceability, with case studies from McDonald's, Training Force, BEE123 and Lulaway.

## Contact

- [Book a 20-minute demo](https://calendly.com/mario-uploader/30min)
- [Log in to Uploader](https://app.uploader.io)
- [Documentation](https://app.uploader.io/docs)

## FAQ: platform

### Is Uploader relevant to us?
If your team tests a high volume of documents against a checklist, yes. Compliance files, audit evidence, onboarding packs, supplier and contractor files, claims or regulatory submissions are all handled with the same method.

### What do you actually check?
Whatever your checklist specifies: that every required document is present, readable and in date, that names, ID numbers and dates match across documents, that forms are signed and certified, and any rules specific to your process.

### What documents can Uploader process?
Any file in any format and language: PDFs, scans, phone photos, handwriting, Word documents, spreadsheets, email and ZIP bundles. Mixed packs are split, every page is classified and named, and duplicates are merged into one record per person or company. There are no templates to build or maintain.

### How do documents get in, and what happens when something is missing?
However they already arrive: portal uploads, email, messaging apps, shared drives or API. Uploader requests missing, expired or incorrect documents from the person who has them and keeps following up. Certificates, permits and medicals are tracked against their expiry dates and flagged before they lapse.

### How is everything checked?
You define your standards once, per document process, country or region, and every file is tested against them, not a sample. Registries, calculations and ledgers are cross-checked where required. Each check returns pass, fail or flagged for review, with the reasoning and source page. Anything that fails goes to a human in the loop, who sees the source page next to the rule, and every decision is logged. Data is extracted once and linked to its source, so it is never retyped across spreadsheets, teams and systems, where manual handovers introduce errors.

### What does Uploader produce?
Clean data captured once and reused for every submission, exported into your systems, dashboards or API, and prepared in each authority's template and coding, and an evidence record your auditor, regulator or funder can review line by line. You can also ask the Uploader agent about any document and receive the answer with its source.

### How long does setup take, and do we need IT?
Most teams run their first live batch within four days. We configure Uploader to your document processes and templates, so no IT department is needed. For regulated environments, Uploader and its AI models can run on premise.

### How established and secure is Uploader?
Uploader processes more than 1.5 million documents a month. A government agency onboarded 27,000 young people through it in a single day. Every view, edit and approval is recorded, and documents can stay inside your own infrastructure.

### How do teams usually start?
With a 20-minute session on your own documents, or a document audit: send any volume as it is and we return a record-by-record report within 72 hours. If you would rather not run it yourself, our full service team completes the packs or clears your backlog for you.

### What does "every page verified" mean?
It means coverage: every page of every file is read and checked against your standard, not a sample. Results Uploader is unsure of go to a person to decide, and every result links back to the page it came from. The auditor, regulator or client still makes the final decision.

### Does Uploader replace our auditor?
No. Uploader does not issue audit opinions, sign assurance reports or carry the auditor’s liability. We deliver the checked, sourced evidence that your auditor, regulator or funder relies on. That evidence has to stand up to an independent party who did not produce it.

## FAQ: SETA, QCTO and YES backlog and 2026 submissions

### When do bookings close?
5 February 2027. That leaves time to obtain missing documents before WSP/ATR is due at midnight on 30 April 2027.

### What kind of backlog can you take on?
SETA, QCTO and related skills submissions built from documents: WSP/ATR, discretionary grants, learnership registrations, QCTO learner and accreditation evidence, YES and Section 12H, with the B-BBEE skills development evidence that supports them. Tell us the programme and we will confirm the requirements.

### Which programmes do you cover?
WSP/ATR (mandatory grant), SETA learner data (SETMIS), discretionary grant claims, learnership registration packs, QCTO learner enrolment, achievement and accreditation evidence, YES youth onboarding and verification, Section 12H schedules and B-BBEE skills development evidence. Running a different programme? Tell us which one and we will confirm the requirements.

### What do you need from us?
Access to the documents as they are, and a contact who can approve packs. We handle sorting, testing and follow-up.

### Can you guarantee we meet the deadline?
For priority bookings (booked before 5 February 2027) we guarantee filing before the deadline, provided the required documents can be obtained in time. Acceptance remains the SETA's decision.

### Who handles our documents?
A dedicated Uploader team works on your backlog under a confidentiality agreement, on our platform, with every view and change logged. Your staff approve the final packs.

### What if a submission is still rejected?
Submissions are most often rejected for missing proof of implementation, missing documents, expired provider accreditation or figures that do not reconcile across forms. Each pack is tested for these before submission. If a rejection is due to our work, we correct and resubmit it at no extra cost.

### Is our data safe?
Yes. Each document is traced to its source, every edit and approval is logged, and nothing is submitted without your approval.

### What happens if we miss 30 April?
The mandatory grant for that year is forfeited; there is no retrospective submission.

## FAQ: use cases

### Do you only work with these industries?
No. If a process runs on documents and has rules to test them against, it can run on Uploader. These are the processes we see most often.

### Which use cases are running in production?
Skills development, HR and payroll, and contractor compliance run for customers today. The rest use the same method with your standards configured.

### Can we start with one process?
Yes. Most teams start with one high-volume process, then add others. Every process draws from the same records.

### Can your team run it for us?
Yes. Our full service team can collect, test and complete the packs for you. You approve each pack.

## Updated

2026-10-08