Services

API & Integration

Most organisations run several software platforms that were never designed to talk to each other. Staff bridge the gaps manually — re-keying orders, copying customer details, reconciling figures between systems at month end. Small Pea Software builds API connections and integration layers that automate data exchange and keep information consistent across your operational stack.

The Case for Integration

Integration work pays off when the same data is entered more than once, when delays between systems cause customer-facing errors, or when reporting requires exporting from multiple sources and merging in spreadsheets. A well-designed integration reduces transcription mistakes, frees staff for higher-value tasks, and gives decision-makers a more current picture of operations.

Not every connection needs real-time synchronisation. Some integrations suit scheduled batch transfers — hourly, nightly, or on demand — while others require immediate event-driven updates when a record changes. We help you choose the pattern that matches business urgency, technical constraints, and the capabilities of the platforms involved.

Typical Integration Scenarios

  • Syncing customers, products, or invoices between a custom application and Xero, MYOB, or QuickBooks.
  • Pulling job or order status from an operational system into a client portal.
  • Connecting a web form or CRM to internal databases and notification services.
  • Feeding completed transactions to warehouse, logistics, or industry-specific platforms.
  • Building middleware that translates between legacy file formats and modern REST or GraphQL APIs.

Integration Challenges We Navigate

Third-party APIs vary enormously in quality, documentation, and reliability. Some vendors rate-limit requests aggressively; others change field definitions without notice. We design integrations with error handling, retry logic, and logging so failures are visible and recoverable rather than silent.

Data mapping is often more complex than clients expect. A "customer" in one system may split into multiple related records elsewhere — contacts, addresses, payment terms, tax settings. We document mappings explicitly and validate edge cases before production traffic flows through the connection.

Our Integration Methodology

We begin by inventorying the systems involved, the direction of data flow, and the authoritative source for each field. That source-of-truth analysis prevents the circular updates and conflicting records that plague poorly planned integrations.

Development proceeds against sandbox or test environments wherever vendors provide them. We build monitoring from the start — structured logs, alert thresholds for repeated failures, and admin screens or reports that show sync status at a glance. You should never wonder whether last night's transfer succeeded.

Security is treated seriously. API keys and credentials are stored securely, access is scoped to minimum necessary permissions, and sensitive payloads use encryption in transit. For integrations handling personal information, we align with your privacy obligations and retention policies.

System integration architecture connecting business platforms

Before You Integrate: A Quick Checklist

Confirm that each platform's API or export capability supports what you need — read the documentation or ask the vendor directly rather than assuming. Identify who owns each system internally and can grant test access. Define what happens when records conflict: which system wins, and who gets notified. Finally, agree on how you will verify correctness during the first weeks of live operation. These steps save considerable rework.

When to Engage Us for API Work

Engage integration specialists when connections are business-critical, when data volumes exceed what manual processes can handle, or when you lack in-house capacity to maintain brittle scripts written years ago. We also help organisations that have built a custom application and now need it connected to the wider ecosystem — a natural phase after core development stabilises.

Our integration projects range from focused one-way feeds to multi-system orchestration with transformation rules, queuing, and administrative tooling. We deliver documentation that explains data flows, failure modes, and how to rotate credentials or adjust mappings when your platforms evolve.

Automated data exchange between connected software systems
Reliable connections that keep data aligned without manual re-entry.

Production integrations fail predictably: duplicate submissions after timeouts or silent stalls on ambiguous vendor errors. Correlation identifiers trace a single transaction across systems when a status never updates.

We document who approves mapping changes and how credential rotation is scheduled so integrations do not become orphaned scripts.

Start a Conversation About Your Connections

If duplicate data entry or delayed synchronisation is costing your team time and accuracy, describe your current setup to Small Pea Software. We will assess feasibility, outline approach options, and provide a realistic view of effort and ongoing maintenance. Contact us or email support@smallpeasoftware.com.

Reliable Integration Patterns in Production

Connections that work in demos often behave differently under production volume and impatient users clicking submit twice. The patterns below address failure modes we see across accounting platforms, CRMs, and operational systems.

Idempotency and Safe Retries

Idempotency ensures submitting the same logical operation twice does not create duplicate records. Stable idempotency keys — derived from internal record identifiers — let receivers return the original outcome on repeat. Retry logic distinguishes transient faults (timeouts, HTTP 503) from permanent ones (validation errors) and documents dead-letter behaviour for operator review.

Webhooks vs Polling

Webhooks push events as they occur — ideal for near-real-time updates when the vendor supports signed, verifiable delivery. Polling queries on a schedule — simpler behind firewalls and easier to throttle, but inherently delayed. We combine both when business risk justifies redundancy: webhooks for speed, scheduled polling for completeness.

  • Webhooks — payment confirmations, inventory reservations, customer-facing status changes.
  • Polling — nightly reconciliation, bulk reference data, immature webhook implementations.

Error Quarantine Queues

Unprocessable messages belong in a quarantine queue — retaining payload, error classification, and correlation ID for deliberate requeue after review — not lost in logs or retried until duplicate side effects appear.

An integration without a visible failure queue trains staff to trust the screen until month-end reconciliation proves otherwise.

Vendor API Lifecycle Management

We monitor deprecation notices, maintain sandbox coverage before production cutovers, and schedule credential rotation and mapping audits as planned activities — not reactions to sudden failures.

Integration Health Reviews

Small Pea Software in Chatswood, NSW provides integration design, remediation, and lifecycle management. Email support@smallpeasoftware.com to discuss an assessment of your existing connections.