ERP stories in 2025 were often AI stories in an ERP wrapper - but not always. That required us to hit the tarmac, and bear down. Old issues with ERP were still with us - including cloud migrations and third-party integration hurdles. Making sense of "agentic ERP" keynote fanfare added to the challenges customers faced - but could ERP vendors deliver?

The Enterprise Stories We Need for 2025

When you look at the new AI tools in application software today, it’s telling how there is no way for users to report these aberrant AI results. The feedback mechanism is non-existent. There’s no way to capture and report the bad AI acts/recommendations. The new AI tech is tucked into an existing process where the reviews/controls/oversight were never part of the process.

**Why? **This article framed five big problems vexing ERP users today. The counsel then is still quite valid at the end of this year. More AI innovations (e.g., agentic AI, small LLMs, etc.) are pouring in faster than companies can assimilate them. Businesses are facing new kinds of technical debt/obsolescence and need faster mechanisms to deeply investigate new tech, mitigate new potential risks, reimagine their business/workflows, and deliver real business value.

Mike Ettling hands over the reins at Unit4, ponders AI’s impact on ERP

I still am an advocate that says transactions still have to be read, written and done, and that’s got to happen in some sort of transactional engine. In the ERP context, what AI is going to do is, everything ERP does for us humans — gives us better data, gives us better insights to make decisions — I think the AI is going to be able to do that faster, quicker, and even without humans. But the transactional piece will still need to happen...

I think the big battle will be, do ERP providers also become provider of all the AI apps and all the AI stuff, or do ERP providers get relegated to a low-cost transactional telco [role]?

Why? 2025 saw a number of tech executives, pundits, etc. opining as to the future of SaaS applications, ERP software, etc. as the AI age unfolds. Sadly, much of this was headline generating click-bait that didn’t really inform software buyers well. However, in Phil Wainewright’s quote from Mike Ettling, we get to the heart of it. The role of ERP in an AI world is being created today but what that holds for ERP vendors is, frankly, up to the affected vendors and the clarity of their visions. *Note: Wainewright also covered a forward view with new Unit4 CEO Simon Paris. *

Workday Innovation Summit review - generative AI fell short for enterprises, but will agents be different?

If I look at what our customers are using from that [gen AI] wave... It’s a lot of nice-to-have features, but it’s not something that’s really consequential on the agentic side. It really is different. If these agents are done right, they’ll act independently; they’ll achieve goals; they’ll learn; they’ll adapt and make decisions. It sounds like a human, and I think that’s the way that we need to think about these agents going forward.

**Why? **Workday’s April 2025 Innovation Summit for analysts brought the issue of generative AI results to a head. As this quote from Workday’s Aneel Bhusri shows, he has a very different take (and hope) for the results of agentic AI in the enterprise than he did for generative AI. The distinction is important. We spent the rest of they year understanding how vendors are approaching agentic AI, and where the opportunities (and risks) lie. But this distinction between gen AI and agentic workflows set the tone for the year.

What does "AI first" cloud ERP actually mean to customers? Acumatica’s CEO and CPO respond

*But unlike so many competitors that just sprinkle a little bit of generative AI on this function, and here’s a little agentic love here, and here’s a little algorithmic kind of thing going on somewhere else, [Acumatica was] demonstrating things where information was coming from everywhere - all over the Acumatica universe - and was populating everything from workflows and report contents and so forth, and directing the customer into things that don’t even resemble a transactional workflow like we used to know - where they were bounded on both sides. *

Why? The AI/ERP debate in 2025 was all about how much AI to embed with ERP and how much AI can SMB users accept. Even the discussions on agentic AI and ERP were looking at whether vendors should take an incremental approach (that helps protect their ERP investments) or go for more radical AI-first re-imaginations of work. 2026 will require more vendors get more specific on this issue as uninspired, piecemeal efforts will not generate great value.

How NetSuite is evolving and the implications for the ERP market

Vendors will need new capabilities, tools, etc. for their new (AI-focused) whole product mix. Specifically, they may need: new process workflows, new controls, new security, significant training materials/courses for users, implementation partners, sales professionals, new skills requirements, and new benchmarks (before and after AI powered apps are implemented).

**Why? **This piece points out just how far NetSuite has progressed, especially after the acquisition by Oracle, and what lies ahead for them to carry their momentum in the AI ERP age. Too many of their competitors haven’t even kitted out the pre-AI solutions well - as few ERP products are truly global, have the same level of applications, etc. This is a reality check for older, smaller ERP vendors on what it will take to be relevant in the ERP AI era.

Certinia 2025 Global Service Dynamics Report - as PSO utilization rates stall, will AI help?

It’s clear from the survey that firms are suffering at the moment from a lack of unified data and automated processes, leaving them far too reliant on spreadsheets and other workarounds to plug the gaps. It therefore makes sense to follow the lead of those higher-margin businesses that have invested in automating their professional services operations and are now investing in their AI strategies. But simply adding AI on top of existing data and processes isn’t going to be the silver bullet that fixes those issues.

Why? PSA (Professional Services Automation) is an important crossover into modern ERP, particularly for service industries, and for as-a-service business models. The same cross-over is true in customer success methodologies, and Certinia has been pushing the conversation forward in both. In this Phil Wainerwright piece, he assess Certinia’s latest PSO report, and examines why PSO firms are struggling, as the promise/challenges of AI disruption add a crucial new twist.

The ASUG Tech Connect AI review - Walter Sun reveals why SAP announced their RPT-1 foundation model, and how it differs from an LLM

I do see two areas of SAP AI that are potentially differentiating. One is end-to-end worklows: I’ll get to that at the end. The other is SAP’s foundation model. As I wrote in SAP makes its agentic AI case, but are customers ready?: ‘But SAP’s foundation model for tabular/structured data is a different pursuit - one that most vendors have not taken on. If this impacts the use of tabular data in SAP’s agentic scenarios, this could shift from a different AI offering to a true AI differentiator, if you get my drift.’

Why? ERP vendors spent most of the year pushing to differentiate on AI. Too often, the messaging was confusingly similar - and not bold enough for our liking. Reality check: most enterprise vendors are pursuing similar strategies for better AI "context" and output (even though they might claim AI uniqueness). I don’t fault vendors for this - responsible/accurate/customer-specific data inputs make LLMs more compelling. Doing that privately and securely has proven approaches now. But as 2025 concluded, we saw more vendors pushing into unique/deeper approaches; something to watch in the new year. SAP’s foundational model advancements were one such example.

Oracle AI World 25 - the AI hot topics and under-the-radar issues you should be watching

At Oracle AI World, the focus sharpened: even with quality enterprise data, LLMs need "annotations." During his analyst day briefing, Juan Loaiza, Oracle EVP of Mission Critical Database Technologies, explained what he meant by annotations for LLMs... What are annotations exactly? You could start with metadata, or perhaps master data. The key: the annotations are telling the LLMs something about the data in the relevant document/database etc. During my talk with Kurapati, he elaborated, with talk of API graphs and metadata.

Why? Oracle was another vendor that sharpened their agentic AI positioning during the fall event season. Oracle also impacted the "AI readiness" conversation. We understand that a big part of enterprise AI results is quality data, but what does that really look like? Graphs? Vectors? Zero copy data layer? Cloud lakehouse? Some of those choices are customer-specific, but Oracle’s clarity on why LLMs need metadata annotations advanced the conversation.

The future of Industrial Operations in an AI age - here’s the IFS vision

*This event was less about press releases, product roadmaps and listing off a litany of product features. It was more about the choices companies have today and what their new technology could deliver. While that statement sounds cliché, the event/messaging wasn’t that at all. No, attendees saw demonstrations of AI, AI-directed robots, smart devices informing operators, operators amending workflows to reflect new knowledge/protocols/etc. and more. *

Why? In the AI ERP world, we’re done with vision statements and theoretical concepts dressed up for a CEO keynote. Software customers and buyers want to touch, feel and experience what AI + ERP will actually do. Not every vendor can do that today. The recent IFS conference showed what was possible today in industrial ERP settings. That marks a turning point in how ERP vendors sell, implement and support customers going forward. The ERP market was already morphing in a big way at the start of 2025 but vendors weren’t exactly moving at the same pace or in the same direction. A ‘solution’ won’t just be some transaction processing code anymore. It must also possess re-imagined workflows, agents, orchestrations, relevant LLMs, great quality data, and much more.