The world of AI is moving at lightning speed. Organizations are busy implementing AI in all their business processes. The amount of profit, cost savings, and new business that organizations can achieve by using AI is unprecedented. Many repetitive tasks are being replaced, processes are becoming increasingly autonomous, all to satisfy customers and generate more business. However, all these different AI technologies also create complexity, because how do you manage all these different AI agents and maintain an overview? That is exactly what ServiceNow wants to offer with its AI Control Tower.
We spoke to Pat Casey, ServiceNow’s CTO and EVP of DevOps, at Knowledge25. We talked to him about what exactly the AI Control Tower has to offer, where it starts and ends, and where the potential lies for the future. The AI Control Tower is certainly not finished yet. It is already usable, and the necessary partners are already connected, but in the future, more visibility and control will undoubtedly be needed and established.
What is the AI Control Tower?
The AI Control Tower is designed to help companies track and manage their growing number of AI technologies. Whether it’s AI agents, the use of LLMs, or even workflows. This is certainly crucial in regulated sectors such as banking, where transparency and control are essential. However, it is also a valuable tool for large organizations that have dozens of AI agents, helping them to maintain an overview.
Features:
- Tracking and Monitoring: The Control Tower tracks all implemented AI agents, monitors their effectiveness, and directs workflows.
- Agent Management: The Control Tower treats AI agents as technological assets, not people (although future applications may change this). The system records details such as calls, accuracy, and other relevant metrics.
- Third-party integration: Although the Control Tower provides detailed data for ServiceNow agents, data for third-party AI agents (such as Microsoft and Google) is currently limited. This is because those organizations do not yet share all log information. Discussions are ongoing.
- Compliance and Governance: The Control Tower helps enforce compliance guidelines, such as using approved LLM vendors and models. AI can also verify that new agents meet these standards.
- Security: The focus is not currently on monitoring data flow but on analyzing prompts for sensitive information. Securing AI models is a new challenge, with prompt injection attacks being a significant concern. ServiceNow is working on patching security vulnerabilities and tracking model defects.
AI Control Tower of the future
The AI Control Tower is still in the early stages of development, but already offers valuable features such as dashboards and data structures. In the future, the AI Control Tower will need to gain even more insight into what AI agents are doing, what data they are processing, and what actions they are taking. This also applies when they are running on third-party platforms. Integration with these third parties will therefore need to be further expanded. Precisely what this will look like remains to be seen.
Overall, the ServiceNow AI Control Tower offers a highly essential solution for managing AI. It is the first of its kind and undoubtedly not the last, but maintaining insight into autonomous processes is and will remain crucial. Although the system is still in its early stages, it already offers many valuable functionalities and promises to play a crucial role in the future of AI management.
Also read: ServiceNow wants to disrupt Salesforce with new AI-based CRM