In September, Syspro launched Torque, a new industrial AI platform developed with manufacturers in mind, particularly those operating on tight margins and needing to know their AI investment would deliver measurable value. The platform is designed to find operational problems, recommend potential fixes and take approved action.
Later that month, at IMTS 2026 in Chicago, Creed Grimm, VP of pre-sales at Syspro, showcased Torque publicly for the first time. Syspro is a mixed-mode manufacturing ERP solution for the mid-market. While the company serves smaller companies doing a few million dollars in business and large operations doing billions of dollars every year, its sweet spot is operators with $10 million to $250 million in revenue. Basically, if you make something by taking materials and ingredients and adding value to them in some way, then you're a good fit for Syspro, Grimm said.
Syspro has been around for nearly 50 years and has a well-established, foundational ERP solution well known in the food and beverage, industrial equipment and fabricated metals industries, among others. Torque adds a new layer of AI capability that works with Syspro's solutions and other software on the market.
"As our customers grow, they may need a full CRM like Salesforce or Syspro’s advanced quality systems," Grimm said. "They find themselves with Syspro and an ancillary database or another solution."
Torque is designed not just to service Syspro, but to pull data from any and all external systems to provide actionable insights.
“When we started looking at the market to see how AI is changing manufacturing, our product team found that there were a lot of answers to that question,” said Grimm
Syspro built Torque to handle package agents (pre-built functions), custom agents (which you control and augment) and enterprise intelligence (which helps request, monitor and solve business challenges).
“It took us a little bit longer to bring this to market than some of the other ERP systems out there, but it's because we strategically and thoughtfully tried to create something that could address as many of those problems as possible,” Grimm added.
Within Torque, users chat with "Sidekick," which interprets tasks and deploys various agents and workflows that pull information from Syspro or external systems, like Salesforce, to execute the challenge.
According to Grimm, every manufacturer is meeting AI technology at a different level. It all comes down to what users are comfortable letting AI do for them. Some start their AI journey with simple "Ask and Act" data retrieval; others with automated, repeatable flows up to governance layers; and some with full trust, where AI independently manages tasks like supplier payments.
Giving automated systems that kind of control can cause some trepidation, so Syspro built in protections.
"Top of mind for a lot of people is AI hallucinations, so our team built what they call the "Glass House," which lets you see everything that's going on," said Jason Adamowicz, product solutions engineer at Syspro.
Syspro’s "Glass House" principle means everything is visible, nothing is hidden. Every AI recommendation shows its reasoning so operators and auditors can follow the logic instead of taking the system’s word for it. The manufacturer decides how much control to delegate, from approving every single action to letting Torque run trusted, repeatable workflows on its own.
When a client picks Syspro, the team first creates connections to different applications, like read-only SQL connections for reporting and Windows Communication Foundation (WCF) connections for writing data properly. Then, they load a knowledge base so Torque understands the company: what it makes, plus specific rules, like not pulling data from the internet. The solution isn't just integrated; it understands the company contextually.
After the connections are made, Syspro deploys the plain-language Torque Builder so non-experts can create agents.
"I can upload a picture of a report with column headers, ask it to build a matching read-only report, and it will confirm the tables, data points and schedule," Adamowicz said. "It even allows you to format the output with HTML for branded email reports with actionable buttons. We can also build agents for things like bill of materials (BOM) intake, where Torque reads an Excel file or PDF, identifies the stock codes and flags any missing data before automatically generating the BOM and operations in the database."
Torque can also monitor external systems. For example, it can monitor an FTP site or email for incoming purchase orders, automatically create an unconfirmed sales order, and draft required raw material purchase orders if inventory is low. Adamowicz said it can also monitor shipping containers, geopolitical landscapes or potential strikes and raise proactive alerts if production schedules might be impacted.
Torque can even interpret ambiguous requests. So, if a user asks Sidekick for the company's "top five orders," it not only answers but explains how it filtered the data, ran the workflow and presents the data in a tabular view where users can ask contextual follow-up questions.
"We’re putting the power in the hands of the users to develop the agents and workflows that they need from start to finish," Adamowicz said. Because "they know their business better than we do."
