If you’re a data integration enthusiast, it’s like being in a race against time to keep up with the AI hype. The world moves at breakneck speed, and it can be overwhelming to keep up with the constant stream of new tools, frameworks, and protocols that promise to revolutionize the industry. But let’s be real, most companies are not startups, and they have to be cautious when it comes to AI adoption.
According to Rogier van Boxtel, Director Pre-Sales Consulting Benelux Nordics DACH, Axway, an enterprise integration and API management provider, the move to AI and MCP requires a lot of governance. This means that organizations need to establish clear policies and guidelines around data integration, security, and compliance, just as they would with any other data integration project. The governance challenge is real, and it’s not just about implementing new tools, but also about ensuring that the AI and MCP systems are aligned with the overall business goals and objectives.
One area that particularly excites van Boxtel is the governance of (agentic) AI. AI is a complex and rapidly evolving field, and governance has always been a topic that arises after a specific integration pattern grows. For instance, we saw the rise of service-oriented architectures (SO
If you’re following the AI space, it’s challenging to keep pace. You probably feel like a lot is happening, and you’re consistently behind the curve. While Silicon Valley startups may be moving at lightning speed, the truth is that most companies must move carefully. This is especially true for innovation at the corporate level.
Rogier van Boxtel, Director Pre-Sales Consulting Benelux Nordics DACH, Axway.
Large organizations must meet strict data compliance regulations and internal guidelines around security and cost control. In general, new tools and patterns require more rigorous testing and validation upfront. This could stall the adoption of cutting-edge AI technologies that hinge on data integration, like Model Context Protocol (MCP), a proposed standard protocol to connect AI agents with data sources, tools, and APIs.
I recently caught up with Rogier van Boxtel, Director Pre-Sales Consulting Benelux Nordics DACH, Axway, an enterprise integration and API management provider, to learn firsthand what AI adoption is looking like within large enterprises. According to Boxtel, the move toward AI and MCP necessitates more governance, just as previous trends in data integration did before.
Read the full interview below for Boxtel’s take on where enterprises are at with MCP, as well as what the future for AI data integration is heading at the corporate level.
Tell me a little about your background working in the API industry. What are some of the latest trends that excite you?
I started working in the integration space in the early 2000s at TIBCO and moved to Axway in early 2014. So, I’ve seen organizations moving from running closed IT infrastructures opened up only for a few trusted business partners to the “opened up” network economy we see today.
The integration software market moved from internal integration to open APIs, and we are now starting to see AI/MCP being added to the mix. If you work in the integration space today, AI is clearly a very interesting and challenging topic.
One area that specifically is of interest to me is the governance of (agentic) AI. Governance has always been a topic that arises after a specific new integration pattern grows: that’s what I saw with service-oriented architectures (SOA) a long time ago, and that is what I saw more recently with API Management. MCP & agentic AI will (again) create a large governance challenge for large organizations, and I’m really curious to see what will happen in that space over the next period of time.
At the Summit, you’ll be speaking on the data integration challenges for corporate AI projects — fine-tuning, RAG, MCP, and governance. From your perspective, where are most enterprises in their AI journeys today?
I’ve always worked with large organizations, ranging from federal government agencies to global enterprises. Those large organizations are really at the tipping point of sandbox-type testing towards business use and value. A lot of activity on the AI side is still in the area of idea to sandboxing — teams are testing out what is possible, what works, and what does not work.
Are you finding common patterns when enterprises begin approaching data integration, especially when connecting with internal or external APIs?
Yes, we do see common patterns. Most companies start with pretty simple point-to-point connections, usually around internal APIs for specific projects. As things grow, they move into more structured API management with governance and security. Many enterprises move from internal enablement to external ecosystems, and once external APIs come into play, like working with partners or opening up services to customers, topics like standardization, scalability, and compliance quickly pop up. Reusability also becomes really important because nobody wants to keep reinventing the wheel.
MCP is still fairly new in practice. Have you seen any promising success stories or adoption patterns yet?
At Axway, we primarily work with large enterprises, and so far, we’ve seen only limited implementations of MCP. Most of the focus has been on testing and running pilots in sandbox environments. We have also adopted MCP within our integration tool, Amplify Fusion, where we are running extensive tests together with customers. As you can imagine, security is a top priority in these scenarios. Gradually, we are starting to see a shift from experimentation toward broader adoption, but it is still very much in the early stages.
What challenges do you see as unique to enterprise use of AI, agents, and emerging protocols like MCP?
For enterprises, the challenge with AI, agents, and MCP is keeping control while still moving fast. They need to stay secure and compliant, but it is often hard to see what is happening inside, costs can spiral, and teams do not always have the right skills. Unlike startups, enterprises can’t just break things — they have to innovate without losing trust or stability.
Once organizations start deploying autonomous AI agents and MCP servers, governance becomes a real issue. How do you see companies avoiding sprawl and managing these systems effectively?
This is already a real topic at one of our enterprise customers, where suddenly multiple MCP servers were spun up across the organization. In our Amplify Engage solution, we are working on functionality to automatically discover MCP servers in the network, so that they can be brought under control. The way forward is to treat MCP servers and agents as first-class assets: register them in a catalog with clear ownership and lifecycle tags. Combine that with central governance for policies like authentication, RBAC, and guardrails.
You’ve participated in many Nordic APIs events in the past. What keeps you and your colleagues coming back?
We keep coming back because it’s a great opportunity to connect with both prospects and existing customers, while also learning from the sessions. It’s a nice and interesting event that’s just the right size, and since we also have an office in Stockholm, for some of our colleagues, it even feels like a home event.
Are there any specific themes or sessions you’re personally excited to check out this year?
I’m personally excited about the sessions on agentic AI and MCP. These are the topics everyone is exploring right now, especially at the enterprise level, which is exactly where we see the biggest challenges and opportunities.
