kognify · Solution for HR & Payroll departments

Autonomous processing of sick-leave notes and medical documents

Dozens and hundreds of HR documents arrive at every larger company's office each month: sick-leave notes, outpatient records, medical board decisions. Kognify processes them autonomously - it reads, validates, links them to the employee, and feeds the finished data into your payroll system. A proven platform, a clear SLA model, a configuration tailored to medical documentation - and zero implementation cost.

Who it's for HR & Payroll department
Platform Kognify Universal Content Handling
Scope Sick-leave · Outpatient · Medical board (LKK/TELK)
Implementation 3–4 weeks to production
The challenge

Sick-leave notes eat into the HR tempo of a large company.

A large company processes hundreds of sick-leave notes every month. Each one is a paper or scanned document that has to be read, linked to a specific employee, validated, and entered into the payroll system before the pay cycle. When the process is manual, it's not only slow but fragile: a wrong date, a missed note, an unrecognized diagnosis - and the monthly payroll comes back for correction.

Today - without Kognify

  • An HR/payroll staffer reviews every document by hand and keys the fields into the system
  • Misread dates or national ID numbers lead to payroll returns and recalculations
  • Identifying the employee by name/ID takes time, especially with duplicate names
  • Documents are kept in folders or scanned without content search
  • A spike in submissions around month-end overloads the team
  • The audit trail is scattered across folders, emails, and Excel

With Kognify Medical Note Handling

  • The document is uploaded (or arrives via integration) and the platform extracts every field automatically
  • Reconciliation against the employee register - the employee is recognized by national ID, name, department
  • Logical checks on dates, periods, and overlaps with other sick-leave notes
  • A structured result, ready for the payroll/HRIS system - JSON, CSV, or direct integration
  • Month-end peaks are absorbed by the system, not the people
  • A full audit trail on every document - when, who, what was changed

The shift: from an HR team that keys in documents to an HR team that only reviews the exceptions and makes decisions.

The solution

Kognify Medical Note Handling - a specialized AI operator for medical documentation.

Built on Kognify Universal Content Handling - the platform already processing documents in production for other clients. A specialized AI operator for medical documentation that lives in a secure environment (Azure EU, SSO with Azure B2C, a separate tenant for your company) and is accessible through an interface and API your team picks up in days.

99% Target success rate for standard, typical sick-leave notes, in line with the other Kognify operators.
3 AI steps with cross-verification - because on sick-leave notes the position of marks and checkboxes carries the core information.
100% Of the implementation, onboarding, and AI optimization is handled by the Kognify team - all the way to the target quality.
What we process

Four types of medical documents - one service.

The service supports the main types of medical documentation. Each type gets its own configuration - which fields to extract, which rules to apply, and which system to send the structured results to.

Type 01 · Core volume

Sick-leave notes

Standardized forms of the National Health Insurance Fund (NHIF) / Ministry of Health.

What's extracted Sick-leave note number, period from/to date, issue date, reason, type, the issuing physician(s)' unique ID, registration number and name/type of the medical facility. Diagnosis by ICD code.
Type 02

Outpatient records

Documents from examinations and consultations. More variable in format - this is where Kognify's logical flexibility comes into play.

We extract Examination date, medical facility, physicians' unique IDs, recommendations for temporary incapacity to work (where applicable). The configuration is open to change through the UI.
Type 03

Medical board (LKK / TELK) decisions

Expert decisions that require specific treatment in payroll - extended absences, reassignment, changed work capacity.

We extract Type of decision, validity period, work restrictions, decision date and number, issuing authority. The configuration is open to change through the UI.
Type 04 · On request

Other HR-relevant documents

Maternity, childcare, medical certificates for operating motor vehicles (relevant for companies with driving and operational staff).

We extract Configurable to the specific type - added when needed, without a separate contract.

GDPR and special categories of personal data: Medical documents contain a special category of personal data under Art. 9 GDPR. Processing is carried out exclusively per the documented instructions of your company (the data controller), in accordance with the DPA signed alongside the contract, with encryption at rest and in transit, and without using the data to train AI models (like all other client data).

How it works

Five steps - from the document arriving to the payroll record.

The same proven Kognify model - only tuned for the HR context. The steps are almost invisible to the end user: you see a finished result and an audit trail, not intermediate states.

1

Document intake

The sick-leave note enters the system through three possible channels: manual upload by an HR operator via the Kognify UI, email (a dedicated address just for sick-leave notes), or integration with the system where employees submit documents (if you have one).

channels: UI · email · API · integration
2

Type recognition

The platform automatically determines whether the document is a sick-leave note, an outpatient record, a medical board (LKK) decision, or something else. This defines which configuration will be applied - which fields to extract and which rules to check.

auto-classification by document type
3

Extraction and validation

The AI operator extracts all relevant fields. Immediately after extraction, logical checks are run: is the national ID valid, does the period make sense, does it overlap with another sick-leave note for the same employee, is the treating physician known in the system.

extraction + multi-layer validation
4

Reconciliation against the employee register

The national ID and names are checked against your company's employee database (via integration or periodic sync). The result: the document is linked to a specific employee record and doesn't stay pending. Unlinked cases go for manual review.

match by national ID + fuzzy fallback by name
5

Delivery to the payroll/HRIS system

The structured result goes where it needs to - direct integration into the payroll system, scheduled export to CSV, or JSON for the next system. Along with it: a link to the original document and a full audit trail on every extracted field.

formats: JSON · CSV · directly to system
Under the hood

What's the process? Three AI steps with cross-verification.

Sick-leave notes aren't ordinary documents - they combine free text, marks, and checkboxes where position carries meaning. That's why we don't rely on a single AI pass, but on a three-stage pipeline in which each subsequent step checks the previous ones with a different context.

01 · Pre-processing

Preparation and context

Turning the file into structurable data. Document-type recognition with custom semi-agentic AI. A preliminary filter of the relevant nomenclatures for the specific type.

  • file → structured data
  • document type recognition
  • custom semi-agentic AI
  • nomenclature pre-filtering
→
02 · Main processing

Extraction with visionAI

Review and extraction of the data as requested for the given document type. Matching against the pre-filtered nomenclatures.

  • visionAI · primary pass
  • type-specific extraction
  • nomenclature matching
  • structured result
→
03 · Post-processing

Verification and cross-check

A second visionAI pass, enriched with the result from main and the data from pre-processing. Compares and verifies across the three sources before the final result.

  • visionAI · verification pass
  • enriched with main + pre context
  • 3-way cross-verification
  • final confidence score

Why three steps and not one? A single AI pass - no matter how good the model - has only one opinion about a document, and even reasoning models have highly variable output. With three specialized steps, each subsequent one checks the previous with a different context, and we keep the process under control. That's the difference between "probably correct" and "soundly verified".

Capabilities

What makes the processing reliable in production.

The difference between a demo and a production system is in the cases that aren't ideal. Here are the mechanisms that make Kognify Medical Note Handling ready for your company's real traffic.

🧠

Logical verification

Not just OCR. The system checks whether the extracted data makes sense - do periods overlap, is the national ID valid, does the diagnosis match the ICD-10 catalog.

🔍

Self-assessment of the result

Each document gets a status: green (ready to send to payroll), yellow (minor uncertainties), red (requires human review). 99% of cases go through the green path.

🔗

Reconciliation with external systems

Integration with the employee register for identification by national ID and name. The option for additional checks against the absence calendar.

⚡

Readiness for peak periods

Month-end, flu season, the periods after holidays - the load spikes. The platform scales automatically, without the HR team noticing.

📋

Full audit trail

For every document: when it arrived, what was extracted, which validations passed, who changed what and when. Ready for HR audits and Labour Inspectorate checks.

🔧

Easy no-code configuration

When a new document type or a new processing rule appears, the configuration is changed through the UI as plain-language instructions - not as an IT project.

Implementation and support

The implementation, onboarding, and AI optimization - handled by the Kognify team.

Kognify takes on the full technical implementation and AI optimization until the target quality is reached. This is not standard industry practice and is part of our vision for the future of AI, where providers will need to take responsibility for the quality of the result so that AI can move toward autonomy.

01
Implementation

Tenant configuration, defining the document types, setting up extraction and validation rules, integration with your systems (payroll / HRIS / employee register).

02
Team onboarding

Training the HR/payroll users on how to work with the UI, how to review exceptions, how to add new rules. Documentation and Q&A sessions.

03
AI optimization

Iterative improvement of extraction quality on your specific documents and formats, until the target success rate is reached. No timer, no separate phases.

Within reason The implementation covers integrating the system specifically for your company (support and changes needed for integrations with payroll/HRIS systems, the employee register, email intake channels), configuring the processing to a satisfactory quality, and changes applicable to your industry or domain (new medical document formats, regulatory changes at the NHIF). Purely custom development and extensive infrastructure/IT/consulting projects that aren't applicable to the industry as a whole are agreed separately as a Change Request under the contract.
Implementation

From signed contract to production in 3 to 4 weeks.

The timeframe is a conservative estimate. With standard, typical documents and available access to the employee register, the real time is often shorter. The Frictionless Start period in the contract can be extended at no additional commitment, if more time is needed for fine-tuning.

Week 1

Setup and document analysis

Activating the tenant for Medical Note Handling. We receive samples of the real document types your company processes and begin configuration. We define which integration is needed to your systems.

Week 2

POC on real documents

A first round of extraction on your real documents (already anonymized, if needed). We review the results together, note errors and specifics, and apply corrections. By the end of the week you have a working POC you can test.

Week 3

Integrations and fine-tuning

Connecting to your systems (payroll, employee register). Setting up the intake channels (email, UI, API). Training the HR team. Fine-tuning on edge cases (non-standard formats, handwritten additions, poor scans).

Week 4

Production launch

Switching to production mode. The first few days - careful monitoring and a fast response to surprises. After stabilization - moving to the usual SLA model.

Why now

Why right now is the moment to automate.

"

Zero risk at the start - we take on the implementation.

The implementation, onboarding, and AI optimization are handled by the Kognify team. You see a working POC on your real documents as early as the second week, and you move to production only once the quality is proven. A clear contract, a valid DPA, premium support - no hidden risks and no long learning curve for the HR/payroll team.

Three things that make the timing right

📈

The economics of volume

Companies with a few hundred sick-leave notes a month are exactly the volume at which automation delivers the fastest effect. A few hundred sick-leave notes, plus outpatient records and medical board documents, take the equivalent of 0.5–1 FTE to process manually - not counting the time spent on corrections during the pay cycle.

🛡️

A regulatory moment

Labour Inspectorate and Revenue Agency checks are becoming ever more thorough. Structured, auditable processing of sick-leave notes with a full trail reduces the risk of penalties during an inspection.

⚙️

Technology maturity

Kognify Universal Content Handling currently runs in production for other clients in similar scenarios. It's not an experiment. The expected results are validated.

Clients and references

Not a demo, but production at some of the largest companies in Bulgaria.

Kognify runs in production for enterprise clients in Bulgaria, Slovenia, and Australia - with a flawless security record and bank-grade audit security. The platform behind this service already processes documents every day for companies whose names you know.

"As one of Bulgaria's leading courier companies, we process thousands of invoices every month. With Kognify, in the very first month of deployment we achieved over 98% successfully processed invoices - while the system was still being tuned. The system independently identifies discrepancies and escalates only genuine exceptions. That gives us confidence without losing control."
TL
Tsvetelina Lazarova
Manager, Projects & Relationships, Finance Group · Speedy AD
Speedy AD uses Kognify for invoice processing and sick-leave note processing - the same service that is the subject of this presentation.

Also trusted by

Konica Minolta Postbank Veolia · Sofia Water TBI Bank Geotechmin Baumax Würth Sopharma PLC VP Brands Minimart Cineland PetPak Slovenia BaptistCare · Australia Bulgarian UN Refugee Agency
Technology partner
of Konica Minolta Bulgaria
Konica Minolta

Ready to start this very month.

A signed contract - and week 1 starts right away. Until then, we can review the document types you process and outline what the implementation looks like at your end.

Get in touch → Meeting to discuss