The Real Growth Story Behind Infor's AI Push
Infor's recent reporting on its own AI progress describes a business growing from both directions of its customer base at once: strong bookings growth of around 30% among its larger enterprise customers, alongside continued momentum at the smaller end of its ERP base where deal sizes are modest but renewal rates stay high. The company says 200 customers have already purchased its newer AI solutions, with what it describes as strong pipeline behind that early cohort heading into the next several quarters of fiscal reporting.
Those are respectable numbers for a vendor whose portfolio includes roughly 60 acquisitions, most made nearly two decades ago, a history that typically leaves a software company with a patchwork of architectures rather than a clean, unified platform. Infor's pitch is that its suites now share a common multi-tenant architecture with unified single sign-on and a consistent user experience across ERP, product lifecycle management, configure-price-quote, and warehouse management, which is precisely what makes rapid AI rollout possible at this pace.
What's in the Stack, and What It's Actually For
The AI stack itself spans three distinct categories: agentic AI for autonomous task execution, generative AI embedded across much of the product line and user interface, and algorithmic AI for the kind of forecasting and optimization work ERP systems have long specialized in doing well. The supporting architecture includes a model context protocol, orchestration agents, headless applications, industry-specific language models, a shared business semantic layer, self-healing agents, retrieval-augmented generation, and compliance agents purpose-built for agent-to-agent interactions across the suite.
In practice, most of the agentic capability so far is built for narrow, discrete use cases rather than long multi-agent workflows that span an entire business process end to end. One unnamed major airline, for instance, used Infor's workforce management tools together with ServiceNow to improve crew scheduling specifically, a bounded and well-defined problem rather than an attempt to automate crew operations broadly. That distinction matters for setting realistic expectations about what current-generation agentic ERP tools can actually deliver this year.
The Multi-Tenant Catch
The architecture that makes this AI rollout possible at such pace is also exactly what limits who can realistically access it today. Customers running older, heavily customized, or single-tenant deployments, even ones technically hosted somewhere described as cloud, have very limited options for adopting these new AI tools without first undertaking a costly architectural upgrade to get onto the shared multi-tenant platform where the capability actually lives and gets maintained. One case cited involves a former client's legacy, on-premises Infor manufacturing system reaching end of life in roughly three and a half years, a timeline that leaves relatively little room for a leisurely, unhurried transition.
This is the detail every CIO evaluating an ERP vendor's AI roadmap should be asking about directly, before signing anything new. The real gatekeeping question is not whether your contract technically includes cloud hosting somewhere in its language. It is whether your specific instance runs on the vendor's current, shared, multi-tenant architecture, because that architectural distinction, far more than the AI subscription tier you ultimately select, determines what you can actually buy, deploy, and get real value from in practice.
Proof Points Are Real, But Still Thin
Infor's examples carry real weight precisely because production AI references in enterprise software remain genuinely rare industry-wide, even now, well past the initial hype cycle. The airline crew scheduling case is a legitimate production deployment running today, not a pilot that quietly stalled after the press release went out, and one scenario the company cites suggests AI could replace up to 90% of human activity in accounts payable invoice processing specifically, though the company is careful to note that figure does not generalize across every function, every industry, or every customer's underlying data quality.
That caveat deserves to be taken seriously rather than read past quickly on the way to the headline figure. AI results in Infor's own account vary considerably by function, driven largely by how repetitive the underlying task is and how clean the organization's data already is going in before any model touches it. Data quality and accessibility problems remain common across the industry, particularly for organizations juggling many spreadsheets alongside multiple overlapping ERP instances, which is exactly the messy environment most mid-market enterprises actually operate in today, not the clean reference architecture used in a demo.
The Bill Nobody's Settled Yet
Infor is bundling AI, process mining, and RPA into a single Velocity Suite specifically to make pricing more predictable for buyers, a response to a real problem: AI consumption and token pricing remain unsettled across the entire enterprise software industry, with no vendor, Infor included, having fully locked down a stable long-term model yet. That instability is not unique to Infor. It is a market-wide condition every CIO signing a multi-year AI contract right now needs to price in as a real risk.
Over half of Infor's implementations run through partners, and roughly a quarter of new bookings come through the channel rather than direct sales, which adds another layer of pricing and scoping variability on top of the base platform economics. Partner-led delivery can speed up implementation, but it also means the final cost and timeline you experience may depend as much on which partner you engage as on Infor's own list pricing.
What to Audit Before You Budget for AI Enablement
Before allocating any budget toward AI enablement on an existing ERP platform, audit your actual tenancy model and customization debt first, rather than taking your contract's hosting terminology at face value. A system labeled cloud in your renewal paperwork can still be single-tenant or heavily customized in ways that quietly exclude you from a vendor's current and future AI roadmap entirely, regardless of what that paperwork technically says about where the underlying servers physically sit.
The second audit item is contractual, and it is easy to skip under deadline pressure: ask your vendor directly what AI consumption pricing looks like today, how it has already changed over the past year, and what protections you have if it changes again before your contract comes up for renewal. Given how openly unsettled this pricing remains across the entire market, treating today's quoted AI costs as fixed for the life of a multi-year agreement is a planning assumption that deserves real scrutiny before anyone builds it into next year's approved budget.
