SINGAPORE — As companies allow artificial intelligence (AI) agents to use their data and applications, data activation company Boomi is offering a central system for deciding what those agents can access, where they can operate, and how much they can spend.
Boomi showcased its Agent Control Plane during a media briefing at the Boomi World Forum in Singapore. The platform brings together tools for building, connecting, monitoring, and governing AI agents, including agents created outside Boomi.
“Every organization has a data problem, an integration problem, an API problem, and now an AI problem,” said David Irecki, CTO and senior director of Solution Engineering for APJ at Boomi.
AI agents can do more than answer questions. They can retrieve company records, update applications, and perform tasks without waiting for a person to handle every step. This also creates risks if companies cannot see which systems the agents are using or stop them from accessing restricted information.
The Agent Control Plane is designed to sit between AI agents and a company’s data and applications. It gives IT teams a single place to monitor agents, apply security rules, and keep records of their activities.
“Once you have that data foundation, you build your agents, whether on Boomi or other platforms, you need a control plane,” Irecki said.
Connectivity is handled through Boomi Connect, which gives authorized AI tools access to more than 1,000 enterprise applications. It uses Model Context Protocol, or MCP, which provides a standard way for AI systems to communicate with external tools and data.
The system follows the access rights assigned to each employee. A worker asking an AI assistant for sales information, for example, should only receive records that the worker is allowed to see.
Boomi’s AI Gateway adds another layer of control. It records interactions involving agents, AI models, and MCP servers while applying identity and security policies to each request. It can also limit usage and send requests to different AI models.
This could help companies manage the cost of using AI. A simple request may be sent to a smaller or locally hosted model, while a more difficult task may be handled by a more powerful external model.
Companies can build workflows through Boomi Orchestrate by describing a business problem in ordinary language. The system can then assemble a proposed solution using integrations, APIs, company data, and AI agents. Boomi Companion similarly lets developers use coding assistants to design and test integrations through conversational instructions.
The control plane also uses Data Hub, Meta Hub, and Knowledge Hub to provide agents with company information and explain what business terms mean. This is intended to reduce inaccurate answers caused by incomplete or poorly understood data.
Boomi is also expanding where companies can run their agents. Its Agent Runtime Clouds support regional deployment, while private-cloud and on-premises options allow sensitive workloads to remain behind a company’s firewall.
“We want you to be able to run AI where you see fit,” Irecki said.