You open a tool you’ve used for years – something you know inside out. And suddenly, it feels slow. Not because it changed, but because you did.
After using AI coding agents; tools that can generate, iterate and solve problems in seconds, your expectations shift. What used to feel fast now feels like friction. What used to feel productive now feels… manual.
This isn’t unique to one tool or workflow. It’s happening everywhere.
And recently, it sparked a candid and thought-provoking discussion in the Safe Software Community Hub. So let’s talk about it.
“It Just Feels Too Slow Now” – A Community Perspective
One long-time FME user shared something that many are quietly thinking.
After nearly 15 years of using FME as their default tool for solving data problems, their workflow has changed dramatically:
“My default tool for random tasks or jobs was FME, now it’s AI agents… Using FME just feels way too slow now by comparison.”
They went on to describe a shift that has become increasingly common:
- AI agents are now the first stop for solving problems
- Tasks that once required building workflows can now be generated instantly
- Even tools like Excel feel new again with AI embedded
- And perhaps most importantly: the role of the “FME expert” feels less certain
The concern wasn’t just about productivity, it was about relevance.
“I’m terrified FME is going to become an expensive legacy tool…”
It’s an honest take, and not an isolated one.
This Isn’t About FME, It’s About A Shift In How We Work
It’s tempting to frame this as a tool comparison:
- FME vs. AI agents, or;
- Visual workflows vs. code generation
But this framing misses the bigger picture: what’s happening is a fundamental shift in how work gets done.
We’re moving from:
- Writing to prompting
- Building to specifying intent
- Iterating manually to letting systems iterate for us
And once you experience an AI agent that can “just do it”, your baseline changes permanently. Speed is no longer a differentiator, it’s an expectation.
AI Isn’t Removing the Work, It’s Moving It
One of the most important insights to come from this discussion is this:
AI isn’t eliminating the need for work. It’s changing where the value lives.
In the past:
- The hard part was writing the code
- The effort was in implementation
Today:
- AI can handle much of the implementation
- The challenge shifts to knowing what to build and how it fits together
This shift elevates new areas of importance:
| Orchestration | How do systems, tools, and workflows connect? |
| Reliability | Can the solution run consistently and at scale? |
| Governance | Is it secure, auditable, and compliant? |
| Data | Can you access, trust, and use the right data in the right context? |
The last point is critical – while AI is incredibly powerful, it still depends on something messy, fragmented and often inaccessible: Real-world Data.
And increasingly, the challenge isn’t just accessing that data – it’s exposing it in a way AI systems can actually use safely and effectively.
This is where emerging standards like Model Context Protocol (MCP) start to play an important role, by providing a structured way for AI systems to interact with tools and data sources.
So Where Does FME Fit in an AI-Driven World?
If AI agents are getting better at generating solutions, where does that leave platforms like FME?
The answer isn’t simply “replaced”, it’s repositioned.
Instead of thinking about FME as a tool for manually building workflows, it’s more useful to think of it as:
- A data access layer connecting systems that AI can’t easily reach
- An orchestration engine for running and managing workflows
- A bridge between AI and real-world data environments
Because here is the reality – AI can generate code, but it still needs:
- Access to on-prem systems
- Secure handling of sensitive data
- Integration across APIs, formats, and infrastructures
That isn’t trivial, and it’s certainly not going away.
The Real Opportunity: AI & FME, Not AI vs. FME
Interestingly, the original community discussion post hinted at the solution:
The desire for an AI agent that can “speak FME”…
This is where things get compelling. Imagine a workflow where:
- You describe the outcome you want
- An AI agent generates an FME workspace
- The workflow is refined automatically
- And it runs reliably across your data ecosystem
In that world, you are not manually building everything, but you’re also not relying on disconnected scripts. You are combining AI’s speed and flexibility with FME’s ability to connect, govern and operationalize data.
That’s not a step-back; that’s an evolution.
What This Means for FME Users
The fear expressed in the community post is real:
“Will I need to find a new career?”
This is being felt across all sectors of the industry. But history is suggesting something different.
Every major shift in tooling – from assembly to high-level languages, from manual coding to frameworks, has changed how people work, not whether they’re needed.
What is changing now?
- Less time spent building from scratch
- More time spent defining, guiding, and validating solutions
This means the role evolves from:
- Builder to Orchestrator
- Executor to Problem Framer
And those who adapt to this shift, especially those who understand both data and AI, will be in a strong position.
FME Will Change, Because It Has To
There’s another important truth here: FME, like every platform, will evolve. It has to.
We’re in the middle of a platform shift driven by AI.
The expectations around speed, usability, and automation have permanently changed.
That means that:
- Interfaces will change
- Workflows will change
- How solutions are created will change
But the core need of connecting and operationalizing data only becomes more important.
The Conversation We Need To Keep Having
What makes this discussion valuable isn’t just the concern, it’s the openness.
These are the kinds of questions that push technology forward:
- What feels slower now?
- What’s becoming easier?
- What no longer makes sense?
- What should exist but doesn’t yet?
The future of tools like FME won’t be shaped in isolation, it will be shaped through conversations like this – by people experimenting, questioning, and sharing what they’re seeing.
So don’t eliminate the need for:
- Reliable execution
- Secure data access
- Scalable orchestration
If anything, they are going to increase. So FME isn’t disappearing in the AI era, it’s becoming something different:
A critical layer that helps AI connect to the real world of data.
And with advancing technologies like MCP, connection becomes more structured, more secure, and more scalable – enabling AI to move beyond isolated tasks and into fully integrated, data-driven workflows.
Whether you are an FME partner, a new or returning user, or a data integration industry professional, make sure to check out our Community website for the latest discussion forums, resources, and information about the latest Safe Software events happening near you.
To find out more about how FME can help solve your data challenges, visit our Platform page.
