The base for successful AI transformation isn’t just adopting new AI tools and assistants. It’s about building a solid and AI-ready data foundation to support agentic architecture and leveraging AI for new business opportunities. An AI-ready data foundation means understanding your context, improving data quality, managing data properly, and designing architectures and integrations that connect systems instead of reinforcing silos. This is the base that enables us to focus on strategic opportunities, or as our colleague Christian Niku says: “With internal friction reduced, focus is liberated from ‘keeping the lights on’ to high-impact business innovation”. Check out Christian’s blog post. You can also read more about AI-ready data from our AI-ready data guide.
The transformation places expectations for different roles and responsibility areas:
- As business leaders, we should consider how we create a safe and enabling environment for everyone to adopt AI, and of course, simultaneously innovating what benefits AI brings to business.
- As developers, we can leverage AI to accelerate time-consuming tasks, build agentic workflows that support platform installations and end-to-end development, and automatically generate integration specifications and business process documentation. This frees up time to focus on what’s important: enabling business value.
At the same time, all of us need resilience to learn and unlearn, adapt to the shifting responsibilities AI brings to our roles and explore how AI can enhance our daily lives.
The key to that is gradual adoption: start with the time-consuming tasks, use AI where it clearly helps, use it to accelerate innovation and build from there. Systematically identify the highest-impact bottlenecks where AI can deliver the most benefit.
One concrete example of transforming the way of working is leveraging AI agents in development, combined with human expertise and insight. Agentic workflows are fundamentally redefining the integration lifecycle and the Integration Developer’s role.
Transforming the Integration Developer role
Advancements in LLMs have made generating integration logic straightforward via prompting. Agentic workflows free Developers from coding minutiae, allowing them to focus on architecture and collaboration. Yet, human expertise remains essential. Integration Specialists manage mission-critical data flows and high-risk environments where architectural control must remain human-led. Understanding data, ensuring robust testing, and validating business processes are responsibilities that cannot be delegated to AI.
While high-level oversight remains a human responsibility, the execution layer is where agentic workflows deliver immediate value. Models are now capable of coding integration flows directly (whether that’s MuleSoft flows with MUnit tests or Apache Camel routes) and generating corresponding test suites. Because these agents can understand the existing codebase contextually, we can significantly improve the codebase through simple agent instructions.
This shift allows teams to automate the tedious aspects of integration development, including:
- Governance & consistency: Agents can enforce project-wide standards for error handling, logging and security patterns across codebases.
- Scalability: Adding capabilities to multiple flows and endpoints with a single prompt, while simultaneously updating the associated tests to accommodate the changes.
- Documentation: Automatically creating or updating documentation after changes occur.
- Quality assurance: Deploying CI/CD agents that review pull requests against specific instruction files.
- Operations & observability: Automating infrastructure provisioning and continuously monitoring integrations for anomalies, enabling proactive incident response and reducing downtime.
However, this automation requires strict guardrails. AI models can hallucinate or introduce security vulnerabilities, especially when handling sensitive data. Therefore, while agents can generate the code, human review remains mandatory for security compliance, data privacy checks, and validating that the logic aligns with organisational policies.
This transforms the Integration Developer from a coder to an orchestrator and business analyst.
Understanding business value and performing analysis have always been integral to integration work, but AI assistance now allows developers to allocate significantly more time to these critical activities. Instead of spending hours on manual coding, developers increasingly focus on assessing business needs and ensuring technical solutions support organisational goals.
The value is no longer in typing the XML or Java code, but in defining the constraints, managing the context, and verifying the outcomes. The developer becomes the architect of the workflow, directing the agents to execute the heavy lifting while ensuring the technical solution aligns with business goals.
We are developing these ways of working together with our customers. That means helping customers build an AI-ready data foundation, find practical ways to use AI agents, and stay clear on where human expertise is still irreplaceable.
If you need help in adopting AI and making your data ready for the change, contact us!
