How do you approach managing multiple enterprise systems (ServiceNow, Workday, Oracle, SAP, Salesforce) that all want to leverage AI? How do you handle security and infrastructure complexity, and do you see a role for an agnostic AI orchestrator?

We have the goal to build an AI integration layer based on the MCP protocol and a central MCP registry and work with companies who are providing enterprise systems to integrate to that MCP-based AI integration layer.

I share the concerns about the hype versus reality of AI benefits. Despite claims of significant productivity gains, many AI projects fail. We are cautious and focus on identifying real, beneficial use cases.

Many enterprise tools already have embedded AI, such as Microsoft Defender and Cisco Umbrella. We have policies requiring legal and IT approval before using any AI tool, prioritizing data confidentiality. While inherent AI tools are widely used, we are cautious about user-driven AI adoption to protect company data.

We view AI in two dimensions: democratized AI (e.g., Copilots, ChatGPT) and strategic AI. Licenses are allocated based on specific criteria to project managers, admin teams, procurement, and legal, where we see good ROI. For strategic AI, we focus on use cases with clear ROI and time-to-market benefits, particularly in engineering and operations. Regarding embedded AI in enterprise systems, our approach is to wait and see, leveraging developments where possible without making strong integration commitments.