IFS has reported that bookings for its Industrial AI products during the first nine months of 2026 were more than three times the total recorded throughout 2025, as manufacturers and other asset-intensive organisations expand their use of industry-specific software.
The company said annual recurring revenue exceeded €1.3 billion at the end of the third quarter, up 24% year on year. Recurring revenue accounted for 85% of total revenue, while more than 100 customers had signed up to pilot IFS Intelligence and its digital workers during the quarter.
IFS has not disclosed the absolute value of Industrial AI bookings in either comparison period. The increase therefore shows the rate of commercial growth against the earlier base, without establishing the value of the underlying contracts or the revenue that has already been recognised from them. The nine-month 2026 figure is also being compared with the whole of 2025.
Customers selecting its Industrial AI technologies during the third quarter included Tesco, LATAM Airlines, Yeo Valley, Mitsubishi Electric and Eight Group. The company identified applications spanning asset lifecycle management, field service, aviation maintenance and manufacturing, but did not publish measured efficiency or reliability improvements for those individual deployments.
The bookings update follows the introduction of IFS Intelligence and its Industrial AI Harness on 13 October. The Harness was announced separately from the 22 October financial results and is intended to give digital workers a controlled route into the applications that manage physical equipment and industrial operations.
The distinction between answering a question and performing an action is important in these environments. A software assistant might retrieve maintenance history or highlight a change in equipment condition. A digital worker could go further, preparing a work order, checking material availability or initiating a permitted step in a service process.
Such actions may affect production schedules, asset availability or activities carried out by maintenance personnel. The software must therefore operate within the permissions, approvals and business rules already governing the relevant process, rather than treating a generated recommendation as sufficient authority to proceed.
IFS says the Industrial AI Harness uses existing identity controls and permissions from connected systems, with planning, orchestration, context management, verification, escalation and audit functions. Its stated design allows a worker to undertake only actions available to the authorised user and to escalate decisions where additional approval is needed.
A work order provides a practical example. An AI system can assemble equipment records and propose a task, but the record must identify the correct asset, the maintenance requirement and the people or resources needed. Approving isolation or returning machinery to service remains subject to the site’s safety procedures and authorised decisions.
The same separation applies to production planning. A scheduling recommendation must account for machine capacity, material availability, changeover requirements and orders already committed to customers. Connecting the software to an enterprise planning application does not guarantee that every proposed schedule is technically feasible.
The Harness includes connectors that determine which systems digital workers can use, together with controls intended to verify actions and preserve an auditable record. The quality of the resulting work still depends on the completeness of asset data, the accuracy of workflows and the customer’s configuration of its permissions.
IFS describes a staged progression from assistance to recommendations and, where approved, autonomous action. That model is more constrained than allowing a general-purpose AI application unrestricted access to industrial systems, but it does not eliminate the need to test a workflow before it is placed into service.
Earlier research published by IFS on 13 October illustrates the gap between adoption and assurance. The company reported that 43% of surveyed frontline teams had trained and introduced agentic digital workers during the previous 12 months, while 11% had validated AI outputs for safety, quality or mission-critical processes.
A separate 11% figure referred to controls for AI use involving sensitive or export-controlled data. These survey findings are not results from the financial quarter and do not measure the proportion of IFS customers with completed production deployments.
Validation depends on the decisions an application is authorised to make. Extracting information for an engineer’s review involves a different level of operational risk from automatically dispatching a crew, changing maintenance priorities or modifying data used in regulated asset records.
Where an AI output has consequences beyond a report, organisations need defined acceptance criteria and an accountable process for correcting errors. Audit logs can record what an application did, but they cannot by themselves prove that its judgement was correct or that the information on which it relied was complete.
Industrial operations also create dependencies between systems. An asset’s maintenance status may affect production planning, spare parts inventory and field service scheduling. A change to one record can influence activity elsewhere, making authorisation boundaries and data integrity important during integration.
IFS is positioning its existing industrial enterprise software as the operating environment in which those controls can be applied. Customers already using its asset management, manufacturing and service applications may have relevant records and approval workflows, although connecting an AI worker still requires configuration and testing.
More than 100 pilot registrations indicate interest in evaluating those capabilities. They do not establish that more than 100 customers have put the software into unrestricted commercial operation, or that the company has independently verified safety and productivity outcomes from those pilots.
The €1.3 billion annual recurring revenue figure similarly describes a revenue run-rate measure across the broader business rather than the value of Industrial AI sales made during the quarter. IFS attributes its growth partly to demand for the technology, without publishing a separate audited contribution from the digital-worker product.
Further evidence of industrial adoption will come from customers completing their evaluations, defining which tasks digital workers may undertake and measuring the results against maintenance, quality and operational requirements. For now, the October update establishes a substantial increase in orders for IFS’s Industrial AI portfolio, alongside an early pilot programme for its newer governed digital-worker capabilities.
