Ask yourself one question.
If an AI agent had direct access to your database and your APIs right now, would it still need your product?
I’ve been building SaaS products for over twenty years. When I ask myself that question honestly, the answer about many of the products I’ve built is uncomfortable.
Not really.
That’s not a technology problem. It’s a design problem.
What Most SaaS Products Actually Are
Strip away the branding. Strip away the landing page copy. Strip away the positioning.
What does the product mechanically do?
Usually, something like this:
A human arrives.
They look at information.
They make a decision.
They click a button.
That’s most SaaS.
A dashboard surfaces information.
A form captures input.
A pipeline moves records from one state to another.
CRUD with a polished UI and a Stripe subscription attached to it.
That model worked for a very long time because humans needed to stay in the loop. They needed interfaces. They needed structure. They needed software to guide them through the process.
But AI changes the economics of workflow software because AI increasingly removes the human from the loop entirely. Large AI labs are already openly discussing agent-based systems that execute tasks directly instead of guiding users through interfaces.
And once that starts happening, you have to ask a hard question:
What part of your product still matters?
Workflow Tools vs Outcome Systems
There’s a distinction I keep coming back to lately.
Workflow tools guide a human through a process.
Outcome systems produce a result.
That sounds subtle until you start applying it to real products.
A support dashboard where a human agent processes tickets is a workflow tool.
A system that reads the support email, checks the order status, determines the refund eligibility, and resolves the issue automatically is an outcome system.
An inventory dashboard showing reorder points is a workflow tool.
A system that analyzes sales velocity and automatically places the purchase order is an outcome system.
The more your product depends on a human showing up and making decisions, the more exposed it is.
The more your product simply delivers the result, the more defensible it becomes.
Most SaaS products are workflow tools.
The uncomfortable part is admitting when your own product falls into that category.
Why Rails Developers Are Especially Exposed
This hits Rails developers particularly hard because Rails is extraordinarily good at building workflow software.
Honestly, that’s one of the reasons Rails became so successful.
The conventions map beautifully onto humans interacting with structured data.
Resources.
Forms.
Dashboards.
CRUD operations.
Admin interfaces.
Pipelines.
I know because I’ve spent years building products exactly like this.
And to be clear, those products worked. One of my SaaS businesses ran for over a decade and served thousands of customers.
But the craft and the market are not the same thing.
Rails optimized many of us toward building software that organizes human workflows efficiently.
I’ve also written about why Rails developers still have a significant advantage in the AI era despite all the panic and doomposting.
AI increasingly rewards software that removes the workflow entirely.
That’s a very different direction.
I ran into this same issue while migrating from OpenSearch to Elasticsearch. The real complexity wasn’t the interface layer. It was the operational logic and infrastructure underneath.
The Real Problem Is Psychological
The technical part is honestly the easy part.
The hard part is psychological.
When something has worked for years, your instinct is to protect it.
You keep shipping features.
You redesign the dashboard.
You improve onboarding.
You stay busy.
Because staying busy lets you avoid the harder question:
Is this still the right thing to build?
That’s what I think is happening across a lot of SaaS right now.
Founders are not blind. Most can feel the shift happening.
But feeling it and acting on it are very different things.
Especially when your current product still pays the bills.
The Defensible Part of Your Product Probably Isn’t the UI
I don’t think every SaaS company disappears.
But I do think a lot of founders are protecting the wrong layer.
The defensible value usually isn’t:
- the dashboard
- the admin panel
- the reporting UI
- the workflow itself
The valuable part is the underlying logic.
The automation.
The data relationships.
The decision-making layer.
The accumulated operational knowledge.
That’s the part AI still needs.
The interface layer is increasingly optional.
That shift matters because many developers still think in terms of building tools humans operate directly.
Meanwhile, the market is slowly moving toward systems that simply produce outcomes.
What I’m Asking Myself Now
The old question was:
Can I build software that helps someone do this faster?
That question built a lot of successful SaaS companies.
Mine included.
The newer question is different:
Can I build a system that removes the need for the workflow entirely?
That’s a different architecture.
A different product strategy.
A different moat.
And honestly, it requires questioning instincts that served many of us extremely well for a long time.
Twenty years into this industry, I’m not asking whether AI threatens my products.
I’m asking whether some of them are still the right thing to build.
Those are very different questions.
One is about defense.
The other is about honesty.