The organizations that navigate market disruption well tend to have one thing in common: they’ve seen the pattern before. We’re watching that pattern play out again right now with AI, and the version of it we learned first, years ago, is worth telling.
About a decade ago, PeopleSoft released a native forms automation tool called Forms and Approval Builder, or FAB. The reaction across the PeopleSoft community was understandable. There was excitement, and a collective exhale. If the platform organizations were already running was about to deliver forms automation, why invest in a third-party solution?
Organizations that had been evaluating GT eForms paused to take a closer look. The logic was reasonable.
The gap between expectation and reality
FAB was designed to do something specific: gather data through a form, route it through a basic workflow, and push it back into PeopleSoft. For simple business processes, it worked.
But most real business processes aren’t that simple. They involve conditional logic, multiple approval paths, exceptions, and updates that touch more than one part of the system. FAB wasn’t built for that. The gap wasn’t in the delivery. It was in the assumption that a built-in forms tool would handle the complexity most organizations were actually dealing with.
One organization spent months trying to make FAB work for processes it was never designed to handle, and invested a meaningful budget with an outside firm in the process, before concluding it couldn’t get them where they needed to go. That experience wasn’t unique. Across the PeopleSoft community, organizations spent significant time and effort discovering, on their own, where those limits were.
Some built substantial custom frameworks around FAB to get it to handle more complex processes, and for a handful of organizations, that effort paid off. But it meant taking on technical work that a purpose-built tool would have handled from the start.
How we approached it
Our position was straightforward, and we held it consistently: if FAB was the right tool for a business process, organizations should use it. We weren’t going to push a product where it wasn’t needed.
But after years of working with PeopleSoft organizations on these kinds of processes, we quickly realized that the situations where FAB was sufficient were the exception, not the rule. So rather than compete on features or wait for the market to draw its own conclusions, we focused on education. We ran conference sessions and built side-by-side comparisons that showed honestly what FAB could do and where its limits were. We hosted workshops, including a well-regarded session at one of the Alliance conferences, that walked through FAB’s capabilities completely. No hard sell, just a clear picture of what the tool was and wasn’t built to do.
Our approach wasn’t about pushing a product. It was about understanding what organizations actually needed and helping them get there.
Organizations invested time trying to make FAB stretch to meet needs it wasn’t designed for, and reasonably so. It’s smart to try what you already have before looking elsewhere. FAB launched in 2014, GT released eForms 3.0 with configurable page creation in 2016, and eForms adoption peaked between 2020 and 2021 as organizations’ needs for reliable automation became unavoidable. Some organizations still use FAB today for the simple use cases it handles well, and that’s exactly as it should be. For organizations with more complex needs, the answer was finding a partner who could actually meet them.
Why this feels familiar
The dynamic we’re watching in the AI market right now is one we’ve seen before.
When a major platform makes a move that looks like it addresses a problem organizations have been trying to solve, the first instinct is to assume the solution is covered. The harder questions get deferred. Will this work for our processes? Our people? Our environment? Organizations move forward on the assumption that what’s available will be sufficient, and the gap between that assumption and reality tends to surface later, after time and effort have already been spent.
That’s the pattern we watched play out with FAB. It’s the pattern we’re watching now. The platforms organizations already run are delivering AI capabilities, and new options are appearing constantly. Tools to evaluate, frameworks to build on, features to turn on. The availability of options isn’t the problem. Knowing which path leads to something that will actually work for a specific organization’s processes and environment is where the difficulty lies, and that’s not something availability solves.
What’s different with AI
With FAB, the gap was a matter of form and workflow features within a defined scope.
AI in an enterprise ERP environment is a different kind of challenge. ERP systems are built to run complex, high-stakes organizational processes. Finance, HR, student administration, operations. These systems carry deep data structures, layered permissions, and years of organizational logic embedded in how they work. Building AI that functions well in that environment means understanding how AI interprets requests, how to structure the underlying data so the AI can use it well, how to anticipate the many different ways people will interact with it, and how to protect against real risks around data access and security. That’s the kind of complexity you learn by doing, across many organizations, many processes, and many implementations. It’s not something you pick up quickly.
We’ve been doing this for a while
Gideon Taylor has been investing in AI for enterprise environments since 2016, long before AI became a mainstream conversation and long before it was something every vendor felt compelled to announce a strategy around.
That’s not a footnote. It’s the reason this moment feels familiar rather than urgent. Being steady when the market is reactive isn’t a posture we adopted for this moment. It’s what happens when you’ve been doing the work long enough to know how these moments tend to resolve.
Where this leaves us
The organizations that navigate moments like this well aren’t always the ones who move fastest. They’re the ones who ask good questions early and find a partner who has done this before, in environments like theirs, with the complexity that comes with them.
Whether you’re evaluating AI options and not sure how to compare them, trying something that isn’t delivering what you expected, or already underway and looking for a more experienced perspective, that’s the conversation we’re built for.
This is the first post in our five-part series, Steady in the Storm: What We’ve Learned About AI in ERP. The next post, Not All AI Is Created Equal, publishes Wednesday, August 19.
