Business Workflow as API: When Simple APIs Aren’t Enough

What if your complex business workflows could be accessed as simple APIs?

Your partner wants to submit purchase orders directly into your system. Simple enough, until you realize the process involves checking three different systems, validating compliance rules, and triggering downstream processes across multiple departments.

This isn’t just about exposing your ERP data. It’s about orchestrating complex business logic across multiple systems and making it accessible as a simple API.

Here’s when you need workflow orchestration, and when you don’t:

Simple APIs vs. Complex Workflows

Simple API (with Management): Customer wants to check order status. Your ERP has this data. But you still need API management for authentication, rate limiting, monitoring, and partner access control. You expose it through an API gateway, not directly from the ERP.

Complex Business Workflow: New order comes in via API. You need to:

• Validate customer credit from your finance system
• Check inventory across multiple warehouses
• Verify compliance requirements from your quality system
• Calculate complex pricing based on contracts and volume
• Route to appropriate fulfillment center
• Update multiple systems with order confirmation
• Trigger supplier notifications if inventory is low

The first scenario needs API management. The second needs workflow orchestration plus API management.

When Your Business Process Needs Workflow Orchestration

Multi-System Data Collection

Scenario: Partner requests current availability for 500+ products. Why workflows help: Data lives in your ERP (financial stock), WMS (physical stock), and production planning system (committed stock). A workflow engine like Kestra can orchestrate the data collection from all three systems, apply business rules, and return consolidated results.

API Pattern: Synchronous if data collection is fast (<30 seconds), asynchronous with status polling for longer processes or callback/notification.

Complex Business Rules

Scenario: Pricing API that considers customer contracts, volume discounts, seasonal promotions, and inventory levels. Why workflows help: Business rules change frequently. Workflow engines let you modify logic without changing API code. You can add new steps, modify calculations, or change approval processes through configuration.

API Pattern: Usually synchronous since pricing calculations are typically fast.

Long-Running Approval Processes

Scenario: Custom order submission that requires engineering approval and production scheduling. Why workflows help: Process might take hours or days. Human approvals, capacity planning, and supplier coordination can’t happen in real-time.

API Pattern: Must be asynchronous. API returns process ID immediately, separate status endpoint for checking progress, webhook notification when complete.

Error Handling and Retry Logic

Scenario: Order processing that involves payment systems, inventory updates, and supplier notifications. Why workflows help: If the payment succeeds but inventory update fails, you need sophisticated rollback and retry logic. Workflow engines handle this automatically with proper state management.

API Pattern: Synchronous for simple validations, asynchronous for multi-step processes with potential failures.

Sync vs Async: Getting the API Pattern Right

Synchronous APIs (seconds to respond):

• Data lookups across a few systems
• Simple calculations and validations
• Quick approval checks
• Real-time inventory queries

Asynchronous APIs (minutes to hours):

• Complex approval workflows with human involvement
• Multi-step processes with external system dependencies
• Batch processing or large data transformations
• Any process that might take longer than 30 seconds
• Fast APIs (milliseconds), but you simply need/want an async approach 🙂

The Pattern: As covered in our Gravitee-Kestra integration post, async workflows need proper correlation IDs, status endpoints, and timeout handling.

Workflow Orchestration Options

When you determine you need workflow orchestration, you have choices:

Complex Business Logic

Best for: Multi-step processes with business rules, human approvals, and complex error handling.

Example: Order-to-cash process with credit checks, inventory allocation, compliance verification, and multi-system updates.

Benefits: Visual workflow design, easy debugging, comprehensive error handling, business-friendly configuration.

Tools: i.e. Kestra

API Pattern: Both sync (for fast workflows) and async (for long-running processes).

Integration Platforms

Best for: High-volume data transformation and system-to-system integration.

Example: Real-time inventory synchronization between ERP and e-commerce platform.

Benefits: High throughput, extensive connector library, mature enterprise patterns.

Tools: ie. Apache Camel

API Pattern: Usually synchronous, optimized for speed.

When to Choose What

• Simple data exposure: API gateway with ERP integration
• Fast multi-system queries: API gateway with simple orchestration
• Complex business processes: Kestra workflows with async API patterns
• High-volume data integration: Integration platform with synchronous APIs
• Long-running processes: Kestra with mandatory async patterns

Real-World Business Workflow Examples

Manufacturing: Custom Order Processing (Async Required)

Challenge: Customer configures custom product via API. Process involves:

• Engineering validation of specifications
• Material availability across suppliers
• Production scheduling optimization
• Quality control planning
• Delivery estimation

Kestra Solution: Long-running workflow with human approvals. API returns order ID immediately, status endpoint tracks progress, webhook notifies completion.

Why Async: Engineering validation might take days, production scheduling depends on current capacity.

Logistics: Shipping Quote (Sync Possible)

Challenge: Customer requests shipping quote via API. Process involves:

• Route optimization across multiple carriers
• Real-time capacity checking
• Dynamic pricing based on current demand

Kestra Solution: Fast workflow queries carrier APIs in parallel, applies business rules, returns optimized options within seconds.

Why Sync: All data lookups are fast, no human intervention required.

Distribution: Automated Replenishment (Async Required)

Challenge: Supplier wants to automate replenishment via API. Process involves:

• Demand forecasting based on historical data
• Seasonal adjustment calculations
• Contract validation and pricing
• Warehouse capacity planning
• Approval workflows for large orders

Kestra Solution: Complex workflow with multiple approval steps. API returns process ID, status updates track approval chain, final webhook confirms order placement.

Why Async: Approval workflows can take days, complex calculations require time.

Technical Architecture: Workflows as APIs

Always Include API Management: Even simple ERP data exposure needs proper API management for security, monitoring, and partner access control. Never expose backend systems directly.

The Pattern: Kestra workflows handle business logic. API gateways manage external access and enforce policies. Message brokers coordinate async communication.

Sync Request Flow:

1. API request arrives at gateway with authentication
2. Gateway validates and routes to workflow
3. Kestra workflow processes business logic quickly (<30 seconds)
4. Results returned directly via API response

Async Request Flow:

1. API request arrives at gateway
2. Gateway triggers workflow and returns process ID immediately
3. Kestra workflow processes in background
4. Client polls status endpoint or receives webhook notification
5. Final results retrieved via separate API call

The key insight

Every business API needs management and control, but only complex processes need workflow orchestration. And long-running workflows usually use async patterns.

Need help designing the right API patterns for your business processes? We’ll assess your workflows and recommend sync vs async approaches, plus the right orchestration strategy.