In ERP as in other areas of enterprise technology, vendors that meet customers where they are, rather than where a roadmap assumes they are, are better positioned to drive real progress. Why? Because while enterprise AI continues to advance, there remains a gap between what is hypothetically, technically possible and what consistently holds up in day-to-day operations. This is why Infor’s approach to agentic ERP starts with industry context and a clear understanding that customers are at different stages of adoption.
Event-driven and agentic capabilities in ERP, which I’ve previously written about, are beginning to appear across vendors, yet consistency and trust in the outputs remain uneven. For the most part, the constraint isn’t the technology. It’s how organizations manage change, align operations, and deal with the friction that legacy ERP continues to create through workarounds, delayed timelines, fragmented data, and a lack of support for real-time decision-making.
Infor is headed into its 2026 analyst summit next week by promoting what it calls “The Agentic Enterprise.” The label is less important than the structure behind it. Infor argues that industry-specific context must precede agentic automation, not follow it. That framing is more in line with how decisions are made inside enterprises and, if it resonates with customers, could better position Infor to support customers at each stage of their AI maturity rather than requiring them to (try to) leap to full autonomy on day one.
In support of its Agentic Enterprise approach, Infor presents a four-stage progression: industry agents, semi-autonomous execution, fully autonomous orchestration, and “governed velocity.” The first stage is where the company has the most traction today, but it wants to move customers along the progression over time.
Before we get to the four stages, let’s consider companies that are trying to embrace AI but haven’t joined this progression at all yet. Generic AI models can produce outputs for them that sound reasonable but miss the key nuances that drive smarter decisions. The example of “waste” makes this clear. If you’re using generic models that lack industry context, then manufacturing scrap, food spoilage, and chemical byproducts may all fall under the “waste” label — yet all of them behave very differently in the real world and require different decisions. Treating them as though they’re the same leads to poor outcomes.
Infor’s response is to build around industry-specific domain language models, supported by process catalogs, a unified data fabric, and a shared semantic layer. In line with this, the company has already introduced industry copilots and tailored workspaces, with more role-based capabilities planned.
That direction makes sense. If a system’s output matches how experienced operators within an industry think, it is more likely to get used. This is where Infor’s depth of industry-specific experience makes a difference. Unlike Infor, some vendors rely on partners to handle the last mile of industry detail, but in my experience that is where gaps can appear. You may get a good generic planning model, for instance, but it may not reflect what is really happening on the floor, such as how scrap, rework, or shift changes affect decisions. Infor builds that context into the platform, which gives it a stronger starting point for producing outputs that people can act on. The challenge is keeping that depth current across industries, regions, and regulations. That requires ongoing investment, but I think the foundation Infor has built positions it to manage that evolution.
The second stage — semi-autonomous — focuses on architecture, where things become more operational. Traditional ERP systems tend to lock workflows and logic into rigid structures that are hard to change and even harder to connect across systems. Infor is moving toward a composable, API-first model that separates the data layer from the application layer. That fits better with how modern supply chains run, where planning, procurement, and financials need to stay coordinated, not siloed.
Organizations running fragmented planning environments often struggle with forecast accuracy and inventory decisions, so aligning data and processes helps improve those outcomes. Infor’s use of ION middleware and low-code configuration is designed to support coordination without requiring a full system replacement. Infor OS includes ION as a standard part of its platform, making it easier for customers to build the connected, composable environments that agentic automation requires. This also eliminates the need for separate middleware charges, effectively treating integration as an add-on cost. Beyond that, the ability to embed into platforms such as Salesforce and support third-party capabilities inside Infor reflects how most enterprise environments operate.
All that said, composability is not free. It only works if process ownership and data definitions are clearly established. Without that discipline, complexity is merely shifted to integration and governance steps, and fragmentation can manifest in different forms.
The third stage is where things become more ambitious. Infor is working toward end-to-end orchestration with what it describes as a “human by exception” approach. It has introduced components such as an Agent Factory and an orchestrator, along with initial use cases, and the company’s roadmap includes additional capabilities such as supervisor agents and cross-platform visibility.
I’ve talked before about the ideas of transforming systems of record into “systems of action.” The upside of achieving that is real, but it also raises the bar for getting decisions right. A flawed dashboard insight is easy to ignore or correct without impacting customers or production. By contrast, a flawed autonomous decision can stop a production line or trigger the wrong order. That is why the earlier, foundational stages matter. Reducing exceptions comes down to two things: First, establishing the industry context so the agent understands how the business actually runs. Second, using a composable platform so the agent can see across systems instead of working with only partial data. If an agent sees inventory but not supplier constraints or production schedules, for example, it could make the wrong call.
Producing better-informed agents should reduce the volume and severity of exceptions. How the remaining exceptions are identified, escalated, and resolved will determine whether organizations trust the model in real-world environments. In my view, Infor’s architecture gives it a stronger starting position than approaches that bolt on industry logic after the fact.
The fourth stage — what Infor calls “governed velocity” — is the most consequential. In this case, governance is not just about maintaining compliance; it is what makes autonomous execution possible at scale. Infor has started building its governance layer that controls how agents operate across systems, ensuring that their actions can be audited. This layer includes centralized identity, zero-standing privilege, continuous authorization, and immutable logging and is expected to have its initial release in October 2026.
That direction makes sense, because it brings into focus the ultimate goal of using agentic AI in ERP: having a system smart enough to make the right calls at speed — and with the appropriate guardrails. The practical challenge, as usual, will be execution. Managing identity, permissions, and policy enforcement across multiple systems and agents adds real complexity. If not handled well, it can slow down the very automation it is meant to enable.
I’ve been able to discuss these developments with Infor’s business and technical leadership on multiple occasions, and for me one of the strongest aspects of the approach is how Infor frames the customer journey as a progression of maturity. This reflects the reality that not every organization is ready for autonomous systems. So Infor maps support across the stages, from customers working exclusively in traditional on-prem environments to more advanced adopters who are ready to implement more of the cutting-edge functionality. Based on my long experience in ERP, Infor’s programs like AI readiness assessments, fixed-fee implementations, and process mining to reduce customization debt show an understanding that many customers are still working through foundational issues.
Moving toward autonomous agentic systems depends on how well customers clean up their data, standardize their processes, and modernize their environments. Infor is not assuming customers are ready on day one. Instead, it supports them through a sensible progression, which shows a more grounded approach to how technology is adopted.
So, we’ve established that the architecture aligns with how enterprises operate. What builds confidence in the field is proof. Measurable improvements in day-to-day operations will matter far more than how complete a vendor’s framework looks on paper.
And then there’s the question of what happens when things go wrong. Multi-agent environments likely won’t break in clean or predictable ways. They’ll conflict, drift, and produce outcomes no one planned for. I believe that Infor’s foundation in industry-specificity and composable architecture should reduce the frequency of those failures, but how the remaining exceptions are detected, escalated, and resolved will make a big difference for Infor’s customers. “Governed velocity” faces the same challenge. The thinking is sound, and Infor has the right structural ingredients in place. But demonstrating that it holds up across real systems will be the next milestone.
Infor can further strengthen its position by putting production case studies in front of the market that demonstrate operational impact from the agentic model. This should include publishing clearer details on how multi-agent failures are handled in live environments, and making the governance layer more tangible for customers still early in their cloud migrations.
All in all, Infor is making a compelling case that agentic ERP depends on industry context, flexible architecture, and built-in governance. Its willingness to meet customers along the maturity journey, combined with structural advantages in industry depth and platform openness, gives it a credible path forward. The next step is demonstrating that these advantages translate into measurable operational outcomes. I’m looking forward to engaging next week at the Infor Innovation Summit 2026 and hearing from Infor and its customers directly on these proof points.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。