Copy and paste as a process
Orders, customer data or documents are copied from one system, amended in a spreadsheet and entered again in the next system.
API integration & workflow automation
API integration and workflow automation for connected operations.
We connect ERP, CRM, finance, operations and third-party systems through reliable APIs and workflows, with traceable failure paths, human approval and clear operational ownership.
Request a workshop See our approach Free first consultation - reply within 24 h.
Where handoffs break
Failures without an owner
An export stalls or a field changes. Nobody sees the failure until a customer, finance or operations team asks what happened.
Automation without operations
The demo workflow runs, but there is no plan for monitoring, access, exceptions, changes or the person responsible for it.
When people become the interface between systems.
Which integration path fits?
Use the standard connector first
If an existing connector covers the flow, data fields and failure handling cleanly, it is often the most economical choice. We check its boundaries and ownership before building custom software.
n8n, Zapier or a workflow platform
Low-code is strong for manageable orchestration and fast, visible workflows. The important questions are not the attractive nodes, but data access, version control, testability, failure paths and who runs the platform day to day.
Custom API integration
For business-critical, data-intensive or rules-heavy workflows, we build a controlled integration service with explicit contracts, traceable state, tests and an exception path that does not depend on luck.
Purpose-built operations software
If the problem is not merely two tools but the entire operating flow, a connector is not enough. We combine integration, business logic and an interface in software that reflects how your team works.
Reliable in production
API and data boundaries
Before the first flow, we establish the source of truth, data ownership, field mapping, validation and permissions. We also define how missing, late or out-of-order data is handled. That keeps it clear where a record originates, where it may change and which version is trusted.
Retries and idempotency
Transient failures need bounded retries with agreed waiting times. Rate limits are respected, and the same business operation must not be created twice. We therefore design retry rules and idempotency together instead of adding them as a late patch.
Monitoring and exceptions
Every run needs a traceable status. Correlation, timestamp, cause and retry history make failures discoverable. Reconciliation also checks whether the expected business outcome actually arrived. Permanent exceptions enter a visible path with an alert and a responsible owner.
Human approval and ownership
Payments, postings, contract data or unusual cases often need domain approval. We make that decision explicit and document who owns operations, access, changes and escalation.
From workshop to operations
Map the process
We map the real flow with systems, people, data, waiting time and exceptions.
Decide the boundaries
Source of truth, access, approvals, cost of failure and operational ownership are established.
Build the core flow
One valuable end-to-end flow goes through development, testing, monitoring and domain acceptance first.
Test in parallel
Where risk requires it, old and new paths run side by side under control, with measurable stop and rollback criteria.
Hand over operations
Documentation, alerts, access, runbook and ownership are handed over or agreed as a managed service.
The first workshop ends with a prioritised core flow and a reasoned tool and architecture decision.
Integration in real operations
All cases
From the magazine
All articles
Integration questions
When is n8n or Zapier enough, and when do we need a custom integration?
A workflow platform suits clear, manageable flows with stable standard connectors and limited failure risk. A custom integration is usually the better fit when the process is business-critical, handles substantial data volumes, needs complex rules or when your team must own its operation and source code. We decide based on process criticality, data, cost of failure and internal ownership, not tool preference.
What happens when a connected system is unavailable?
Predictable failure is part of the design. Transient errors are retried in a controlled way, permanent failures enter a visible exception path, and critical cases are handed to a responsible person. A failure must not silently lose data or execute the same business operation twice.
How do you prevent duplicate orders, postings or emails?
Business-critical steps are designed to be idempotent. The same operation can arrive again without creating a second business outcome. Depending on the systems, we use stable operation identifiers, idempotency keys, state checks or controlled deduplication.
Can you connect legacy systems without a modern API?
Often, yes. Viable paths include file imports, approved database access, email or document workflows and small adapter services. First, we establish which system remains the source of truth and which access method is acceptable from business, security and operational perspectives.
Who operates the automation after launch?
We agree this before development starts. Responsibilities for operations, monitoring, exception handling, access and changes are documented to the agreed scope. DUNA can operate the integration or prepare your team to own it internally. The proposal states which source code, workflow configurations, access rights and documentation are included in the handover for the chosen solution.
How do you handle sensitive data and system access?
We limit access to what the workflow needs, separate test and production, and document which data crosses each system boundary. Credentials belong in suitable secret management, not visible workflow fields or source code. Data location, retention and deletion are agreed for the specific process with the responsible stakeholders.
What do we get from the automation and integration workshop?
We map the current flow, establish systems and data ownership, prioritise broken handoffs and risks, and select one useful end-to-end flow to tackle first. You leave with a grounded recommendation for a standard connector, workflow platform, custom integration or purpose-built operations software, plus a clear next step.
Request a workshop
Where does the handoff break? Show us.
Tell us the systems, the handoff and the most visible problem. We will reply within 24 hours with an honest assessment of the most useful first step.
Prefer to talk? +49 173 6618059