Manufacturers back PLM as top data source for AI agents

Propel has published research showing that manufacturing executives rank product lifecycle management systems as the top data source for AI agents using model context protocol. The survey covered 400 senior manufacturing professionals.

The findings suggest manufacturers increasingly see product data as the main requirement for using AI tools in business processes. In the study, conducted by Talker Research for Propel, 49% of respondents ranked PLM as the top business system for AI agents to access through MCP, while 38% said PLM provides the most valuable data for AI-driven actions.

MCP, or model context protocol, is an open standard designed to let AI systems connect with and act on data held in business software. The survey focused on manufacturers in high tech and electronics, industrial equipment, and medical devices, and included respondents from management and board level to C-suite roles.

The results point to a broader shift in how senior manufacturing leaders assess AI projects. Rather than treating AI adoption as a standalone software decision, respondents tied its value to the quality and accessibility of product-related information, including requirements, bills of materials, and engineering change records.

Nearly nine in 10 respondents, or 89%, said a connected product data foundation is essential to realising AI's potential. Another 75% said it is critical or very important for AI systems to understand links across product, quality, supplier, manufacturing, and customer information.

The survey also found that the absence of product data would undermine many AI efforts. Some 66% of respondents said AI would be completely or mostly ineffective in their organisation if it could not access product-related data.

MCP uptake

Interest in MCP appears to be spreading quickly across senior leadership teams. Overall, 91% of respondents reported increased organisational interest in MCP, with 63% describing that interest as dramatic or significant.

Among the most senior respondents, 96% of CxOs and 85% of board members said interest had risen. That suggests the issue has moved beyond technical teams into wider business planning, particularly as manufacturers look for ways to connect AI systems to existing operational data.

More than 90% of respondents said they have already implemented MCP or expect to do so within the next year. A further 56% said delaying adoption would lead to higher operating costs or a competitive disadvantage.

That expectation of near-term change was matched by longer-term views on business impact. Eighty-five percent of senior leaders said MCP would be transformational within three years, while executive vice presidents and vice presidents estimated that nearly one-third of their daily business processes would be automated or influenced by MCP.

Product record

The research adds weight to a debate already under way in manufacturing software over which systems should sit closest to AI decision-making. Enterprise resource planning, customer systems, and quality tools all hold operational information, but respondents placed PLM first when asked where AI agents should draw context and action data.

Ross Meyercord, Chief Executive Officer at Propel, said the reason lies in the reliability of the underlying record. "MCP gives AI systems a way to access data, but reach is only as valuable as the data it finds. If your product record is fragmented, outdated, or siloed, you've just given AI a faster path to bad answers. PLM solves that. It's the system that keeps product knowledge current, connected, and trustworthy," he said.

He added that manufacturers appear to be preparing for MCP with that concern in mind. "This research demonstrates that manufacturers already understand the stakes and they're planning for MCP to bridge the gap between AI and product data," Meyercord said.

The results come as manufacturers face pressure to use AI not only for content generation or search, but also for workflow decisions tied to design, sourcing, quality, and service. In those settings, incomplete or inconsistent product information can create operational risks, especially in regulated sectors such as medical devices and in complex industrial supply chains.

By highlighting PLM's role, the survey also reflects a broader effort by software suppliers to position product data systems at the centre of AI strategy. For manufacturing groups, the practical question may be less whether to adopt AI interfaces and more whether the underlying product record is structured well enough for those systems to use.

For the executives surveyed, that appears to be the central issue: AI access matters, but trusted product data matters more.