Orchestrate.Document Automation.Your Way.
Document automation software that adapts to your business logic. Build intelligent document processing pipelines with branching, parallel processing, and seamless integrations.
One Pipeline. Many Possibilities.
MiruIQ pipelines are built from modular components that can be freely combined, extended, and adapted: document workflow automation without rigid templates.
You define how documents should flow; MiruIQ handles execution, routing, and coordination across systems.
Composable
Combine modules freely (classifiers, extractors, verification steps, and connectors) in an order that fits your process. Pipelines are designed to evolve: add, remove, or replace steps without redesigning the entire flow.
Sequential & Parallel
Execute steps in order when results depend on each other, or run tasks in parallel when they don't. This allows routing files, extracting data, and triggering downstream actions at the same time, within a single pipeline.
Branching Logic
Route documents to different paths based on classification results or extracted values. Each branch can apply its own extraction rules, verification logic, and delivery targets, all within the same pipeline.
Multiple Outcomes
A single pipeline can produce multiple results at once. Documents can be stored, forwarded, or re-classified while structured data is delivered to databases, APIs, or automation systems.
Automated Document Routing
See how a property management company handles incoming documents from tenants, vendors, and service providers.
Every day, documents arrive from multiple sources: emails, cloud folders, and API uploads. The pipeline automatically identifies each document type, extracts relevant data, and routes it to the appropriate system or workflow: automated data entry that replaces manual retyping.
Centralized Intake
Documents arrive from email inboxes, cloud storage, and API webhooks into a single processing queue.
Document Identification
A classifier determines whether each document is an invoice, contract, maintenance request, or other type.
Parallel Extraction
Each document type triggers a specialized extractor that knows exactly which fields to capture.
Automated Routing
Based on document type and extracted data, documents are routed to accounting, legal, or operations.
Workflow Handoff
Verification modules ensure data quality before passing to downstream systems like ERP or DMS.
Structured Outcome
Clean, validated data flows to the right destination: database, webhook, or external system.
Loan Application Processing
See how a lender processes multi-document loan applications with cross-document verification.
Loan applications arrive as document bundles containing ID documents, income statements, and bank statements. The pipeline separates, extracts, and cross-validates information across all documents before making a routing decision.
Application Intake
A document bundle arrives containing all required materials for a loan application.
Document Identification
A multi-structure classifier identifies each document within the bundle: ID, payslips, bank statements.
Parallel Extraction
Specialized extractors run simultaneously on each document type, maximizing throughput.
Application Assembly
Extracted data from all documents is aggregated into a unified application record.
Cross-Document Verification
Verification modules check consistency: does the name on the ID match the payslip? Do income figures align?
Final Routing
Based on verification results, applications route to auto-approval or manual review queues.
How Pipelines Are Composed
Pipelines are built from three fundamental patterns that can be combined freely into any document processing workflow.
These patterns can be nested and combined to create sophisticated document processing workflows that match your exact business requirements.
Sequential Steps
Modules execute in order, each receiving the output of the previous step. Use for linear workflows where each step depends on the last.
Parallel Paths
Multiple modules execute simultaneously, then merge results. Use when independent operations can run concurrently.
Branching Logic
Conditional routing based on classification or data values. Use when different document types need different handling.
Create Pipelines Your Way
Build pipelines visually, programmatically via API, or through AI-powered MCP tools.
Visual Pipeline Editor
Build and trigger pipelines visually
The visual pipeline editor lets you design, understand, and evolve document pipelines without writing code. Sources, modules, and sinks are connected into a clear flow that reflects how documents actually move through your business.
An operations team builds a pipeline visually to process incoming documents. They connect a storage source to classifiers, route documents to different extractors, and deliver results to multiple systems. New document types are added later by extending the pipeline, without redeploying applications.
API-Driven Pipelines
Trigger pipelines programmatically
API-driven pipelines allow applications to interact with MiruIQ directly. Files, signals, and parameters can be sent into a pipeline, which then executes its full processing logic automatically.
A loan management system uploads documents into a MiruIQ pipeline via API. Each uploaded file flows through classification, extraction, and verification steps automatically. Once processing completes, structured data is delivered back to internal systems while the original files are routed to storage.
MCP-Driven Pipelines
AI-initiated automation
MCP-driven pipelines allow AI tools and agents to initiate structured document automation. Instead of treating AI output as an endpoint, MiruIQ turns it into a controlled, auditable processing flow.
An AI assistant gathers information from external sources and generates a document. Using MCP, the assistant uploads the file into a MiruIQ pipeline. The pipeline classifies the document, extracts structured data, and triggers downstream systems, turning AI output into controlled enterprise automation.
Design pipelines visually with drag-and-drop components.
Connect Where Your Data Lives
MiruIQ pipelines don't require you to move documents into a specific system or format first.
Connectors allow pipelines to ingest documents from existing systems and deliver results directly to where they are needed, without intermediate storage or custom glue code.
Sources and sinks are first-class pipeline components. They can be combined, branched, and reused across pipelines to reflect real operational flows.
Source Connectors
Bring documents into a pipeline
Source connectors define how documents enter a pipeline. They support both event-driven ingestion and explicit triggering, allowing pipelines to start automatically or on demand. Whether documents arrive one at a time or as bulk drops, the same pipeline handles event-driven and batch document processing alike.
A document is uploaded via an API endpoint into a pipeline. Immediately after ingestion, the pipeline classifies the document and routes it to the correct processing path, without the application needing to know what happens next.
Sink Connectors
Deliver results and trigger action
Sink connectors define where pipeline outputs go. They handle both structured data delivery and file routing, and can also trigger further automation once a pipeline step completes.
After extraction, structured data is written to a database for analytics. In parallel, the original file is stored in a secure archive, and a webhook notifies another system to continue processing, all within the same pipeline.
Designed for Reuse and Composition
Connectors are decoupled from pipeline logic. The same source can feed multiple pipelines, and the same sink can receive outputs from different processing paths.
This allows teams to standardize ingestion points, reuse delivery mechanisms, and evolve pipelines without re-integrating systems.
Common Questions About Pipelines
Short answers to the questions teams ask before automating their document workflows.
Document processing automation means documents are classified, extracted, verified, and delivered to downstream systems without manual handling. In MiruIQ, this is modeled as a pipeline: documents enter through source connectors, flow through classifier, extractor, and verification modules, and leave as clean structured data, with human review only where you decide it is needed.
Each pipeline is composed from modular steps that mirror your real process: sequential steps for dependent logic, parallel paths for independent work, and branching for routing decisions. Because document workflow automation is defined by structure and meaning instead of layout, pipelines keep working when document formats change.
Yes. MiruIQ runs as a managed service on Swiss infrastructure or fully on-premises in your own environment, including the AI models. Documents never have to leave your infrastructure: no US cloud dependency, no data leaving Switzerland in the managed setup.
Four things: ingestion from where documents actually arrive (email, S3, SFTP, API), extraction that survives your hardest documents, human review built into the flow, and deployment that matches your data constraints. Before trusting any 'document processing automation top picks' list, run your five hardest documents through the candidates.
Automation That Grows With You
Start with a simple pipeline and scale document processing automation to enterprise complexity, without rebuilding from scratch.