Classify and Split Documents Automatically
Turn mixed page streams into structured, auditable documents using deterministic rules and governed AI. Deploy on-prem, private cloud, or SaaS—without changing your downstream workflows.
Stop manual sorting and costly mis-splits
Manual intake breaks down when documents arrive as mixed PDFs, scanned batches, and email attachments. Teams lose time separating pages, guessing document types, and fixing boundary errors—creating downstream exceptions, compliance risk, and unpredictable SLAs.
IRIS Classification & Splitting applies a governed, hybrid approach: deterministic separation (barcodes, patch codes, value-change rules, fingerprints) where you need speed and certainty, and AI-driven classification where variability is high. Low-confidence cases are routed to verification so every decision can be corrected, tracked, and continuously improved.
Multi-channel intake
Classify and split documents from scanners, folders, email, and cloud sources in one consistent pipeline.
Deterministic separation
Use barcodes, patch codes, blank-page detection, and value-change rules for precise boundaries at speed.
Fingerprint-based identification
Recognize stable layouts to classify pages and trigger reliable splitting without fragile templates.
Hybrid AI classification
Apply AI models for high-variance document sets while preserving operational governance and traceability.
Human-in-the-loop verification
Route exceptions to review queues to prevent errors from reaching ERP/ECM systems.
Audit-ready decisions
Capture classification results, confidence, and interventions to support compliance and chain-of-custody.
Operational telemetry
Track KPIs like throughput, auto-classified, exception rate, and verification time for SLA control.
Integration-first outputs
Export structured documents and metadata to ECM, ERP, RPA, and custom apps with predictable payloads.
Flexible deployment models
Run at the edge (desktop), in the core (server), or as cloud microservices—without rewriting processes.
Portfolio-wide capability
Operates across IRISPowerscan, IRISXtract, IRISPulse, and SDK engines with a consistent approach
Designed for hybrid workflows, APIs, and governance
IRIS implements Classification & Splitting as a horizontal capability across desktop capture, server platforms, cloud microservices, and SDKs—so you can standardize intake logic and reuse it across teams and channels. Deterministic methods (barcodes/patch codes/fingerprints/value-change expressions) deliver low-latency, high-confidence separation, while AI engines (e.g., NClassify/XClassify-based approaches) handle long-tail variability with confidence scoring and controlled thresholds.
For modern architectures, IRIS services integrate through REST APIs and SDK patterns, and in cloud deployments results are carried through queue-based pipelines with companion metadata—making every classification and split decision inspectable, debug-able, and audit-friendly. This enables “rules → AI → verify” orchestration without turning AI into a black box, uncontrolled.
Your Questions Answered
Is this ‘AI-only’, or can we keep deterministic control?
You can start fully deterministic (barcodes, patch codes, fingerprints) and layer AI only where variance demands it. Hybrid orchestration keeps outcomes explainable and operationally governed.
What happens when classification confidence is low?
Low-confidence documents can be routed to verification queues for human review. Corrections can feed continuous improvement, while preventing bad data from reaching ERP/ECM.
Can it handle email attachments and mixed PDFs, not just scans?
Yes. The capability is designed for multi-channel ingestion (scan, email, folders, cloud) and applies the same classification/splitting logic across sources.
Do we need to redesign our downstream workflow to use it?
No. Outputs are designed to integrate into existing export and validation flows. The goal is to improve intake structure while preserving downstream systems and controls.
Can we deploy on-prem for data residency requirements?
Yes. IRIS supports on-prem, private cloud, SaaS, and hybrid models—so you can align deployment to compliance, security, and operational constraints.
A practical path to governed automation
Start deterministic, add AI where it adds value, and keep humans for outliers—measurably.
Assess document families
Identify top inbound flows (mailroom, AP, HR/legal) and define target document classes and boundary rules.
Configure deterministic separation
Apply barcodes/patch codes, fingerprints, blank detection, and value-change rules for fast, reliable splitting.
Introduce AI classification
Use AI models for high-variance sets; tune thresholds and confidence routing to minimize exceptions.
Verify and govern
Route exceptions to review, capture audit trails, and feed corrections back into continuous improvement.
Measure and scale
Monitor KPIs (throughput, % auto-classified, exception rate) and scale services or stations as volume grows.
Make intake faster and correct
See how hybrid rules + AI improves throughput without losing control.