If you are responsible for managing the “built environment”, covering the physical world, any assets in that world and the physical spaces they inhabit, you may be thinking that you have a little more time to prepare for the impact of gen AI and agentic AI than those managing business workflows and processes in the front and back offices.
However, Accruent, the Texan based provider of SaaS software to manage physical resources and facilities, begs to differ. It is committed to helping customers be prepared for gen AI and agentic AI era by providing Model Context Protocol (MCP) apps. As the company is preparing for the launch of these apps, I spoke with Chief Product and Technology Officer at Accruent, Aron England, to discuss this market opportunity.
From spreadsheets to gen AI
Accruent works with all types of industries and its customers range from micro-businesses such as dental offices up to well-known Fortune 500 retail brands. It works with regional services partners, and has recently signed a new deal with Cognizant to implement and help optimise Accruent’s software alongside other systems that customers are using.
Thinking about the range of customers served, England comments:
We have met with customers who are managing assets using spreadsheets and an issue there is that you need to find the right spreadsheet to update and then frequently people accidentally over-write information. With our system you get automation and a single source of truth.
Accruent’s maintenance management system, Maintenance Connection, is used by Toshiba International Corporation at its facility in Houston to manage facilities, assets, work orders, and to dispatch parts. England explains:
We serve over 100 users a month, transforming current manual tasks into what is happening in the history of an asset to manage repetitive breakdowns. The benefit is that they are moving from no visibility and no understanding of asset maintenance to having the information necessary to justify their headcount and operational expense. Before they did not have all the notes about the assets. The system is not preventative but it can identify patterns of breakdown. There is access to better data quality through our system because we can accurately record hours and notes as well as using Gen AI to access vendor management notes.
Moving to preventive maintenance with Accruent MCP apps
As many vendors do in this market, Accruent has machine learning models and RPA in its products and it has also integrated Gen AI into the portfolio. It is poised to go further as England indicates:
Later this year we will be introducing a full agentic AI platform so that our engineering teams and next year, our partners and customers will be able to build asset management agents and orchestrate them. With gen AI a user has to have pre-meditated intent and hope they get the right response back. With the new platform you can create one or many agents in a supervised or unsupervised manner that can work autonomously as MCP apps to amicably build end user experiences.
One example of how this will work is that for asset management, a set of agents will go through an asset on a daily basis and open or close work orders, and pull vendor documents and maintenance schedules together to expedite a work order on behalf of the user.
For some customers the demand for this type of system comes from the scale of their environment. For example, Accruent has an airport customer with 200,000 assets and it is virtually impossible for humans to manage maintenance cycles for an estate of this size. Other customers are also struggling with preventive maintenance. They have implemented a solution for corrective maintenance and need to invest more to create preventive maintenance and this is where Accruent believes its agentic platform comes in. England continues:
The approach goes beyond assets. We are also exploring AI data extraction from documents. Think about retail leases. Going through these is very time-consuming but the information is critical for dictating financial responsibility. For example, a customer may get a new snow plough delivered that costs $80,000, or a new refrigeration pump for $125,000 and they do not know if they are financially responsible for this or not. If we can save them these kinds of dollar sums, that will go straight to the bottom line. Our tool links back to the original vendor documentation via citation, so the human in the loop can double-check.
Aside from using MCP for content, the AI platform also has guardrails based on the customer’s own precision tooling and will not leak data because only those with the organization’s access permissions can use it.
Accruent plans to go further and create foundation models for specific use cases. England says:
In Q3 we will also be launching a predictive failure system for retailers’ refrigeration systems. We are taking data across customer IoT sensor data to build the foundational model, and then we can build models tuned for individual customer environments on top of that. These bespoke models are designed for their facilities and operations, tied to their uniqueness. The models will look for anomalies in the data coming from real-time streams and generate work orders and get them dispatched. When you are working with very thin margins, this type of system can make a big difference to profitability.
We believe that we are ahead of the market as we enter a new era, transitioning from RPA (Robotic Process Automation) and Machine Learning models that are brittle and expensive to deploying AI into products and customer systems in days, weeks or hours.
My take
In the operational technology area, the market is at a point where organizations are looking to replace older technology by moving to cloud and SaaS. This should offer Accruent and its partners the ability to suggest moving to an AI maintenance management platform at the same time. A wider market issue for asset manufacturers to consider is the impact such maintenance systems will have on extending the lifetimes of assets and what that might mean for their business models.