Do You Still Need Zapier in the Age of AI?

For years, whenever I needed to connect two applications, my instinct was simple.

“There must be a Zapier integration for that.”

If there weren’t, I’d look for Make (formerly Integromat), and before that I’d even experimented with IFTTT for simpler consumer and smart home automations. More recently, platforms such as n8n have expanded the choices available even further. It wasn’t that I needed IFTTT particularly often, but like many people interested in automation, I was fascinated by what could be connected. Zapier became my first serious automation platform; Make offered greater flexibility as my workflows became more sophisticated, and together they fundamentally changed how I approached software.

At the time, that progression made perfect sense.

As cloud software exploded, integration platforms solved a genuine problem. They allowed businesses to automate repetitive tasks, move data between applications, and build surprisingly capable workflows free of writing code. I built some fairly complex automations over the years, and for a long time I wouldn’t have wanted to be without them.

Today, however, something has changed.

Not because Zapier, Make, or any of these platforms have gotten worse. Quite the opposite, they’ve continued to improve. The difference is that my own workflow has evolved, and with it, the first question I ask whenever I need to connect two systems.

A year ago, my thought process was usually quite simple. I needed to connect two applications, so I immediately began looking for the best integration platform to bridge the gap.

Today, my instinct is very different.

Before I even consider another subscription, I find myself asking a much simpler question.

Can I just build this instead?

That shift has happened remarkably quickly.

Platforms such as Lovable and Emergent, combined with increasingly capable AI models, have radically transformed what’s possible for founders, small businesses, and even solo developers. Rather than opening an integration platform and designing another workflow, I now find myself exposing an API endpoint, explaining what I want to achieve in plain English, and allowing AI to generate the logic that connects everything.

The irony is that I never consciously decided to replace integration platforms. In fact, I was a happy Zapier user for years before moving increasingly complicated workflows into Make. It was only while rebuilding projects such as Slick Media using AI-assisted development that I realised my behaviour had quietly changed. When I need a new integration today, I don’t instinctively search for another platform. I open Lovable or Emergent, describe the outcome I’m trying to achieve, expose an API endpoint if necessary, and begin building. The transition wasn’t deliberate; it simply became the quickest route from an idea to a working solution.

That doesn’t mean I never use integration platforms anymore. It simply means they are no longer my starting point. They’ve become a fallback rather than the obvious first choice. The important point isn’t that Zapier has become less capable. It’s that AI has made me more capable.

The Economics Have Changed

Perhaps the biggest surprise has been how quickly the economics have shifted. Integration platforms didn’t just save time; they justified another monthly subscription because the alternative often meant writing custom code or hiring a developer. That calculation now looks very different. If I’m already paying for AI, the cost of creating a bespoke integration has fallen dramatically. Many workflows that would previously have justified another SaaS subscription can now be built in an afternoon, tailored precisely to my own requirements.

That doesn’t make integration platforms poor value, far from it. It simply raises the bar for when I decide another recurring subscription is genuinely worthwhile.

That shift’s affected something else I wasn’t expecting. It has fundamentally changed how I buy software. For years, the process was relatively straightforward. You identified a problem, searched for software that solved it, compared features and pricing, then added another subscription to your monthly outgoings. Increasingly, there’s a new question that comes before all of that.

Should I buy software for this, or can AI build enough of what I actually need?

That doesn’t mean AI replaces SaaS, nor does it mean every business should start building bespoke applications. What it does mean is that every software purchase now has another competitor.

Not another SaaS platform. A blank prompt.

APIs Have Become My New Buying Criteria

This has also changed how I evaluate software itself. A few years ago, seeing “Works with Zapier” on a product page was genuinely exciting. It immediately suggested flexibility and reassured me that I could connect the product into the rest of my workflow. Today, that badge carries far less weight. Instead, my attention goes entirely elsewhere.

• Does the platform have a well-documented API?
• Does it support webhooks?
• Is authentication straightforward?
• Can AI understand the documentation well enough to build against it?

Increasingly, I judge software by the quality of its API rather than the size of its integration marketplace.

A clean, well-documented API tells me the platform is extensible. More importantly, it tells me AI has an excellent chance of building exactly that integration I need without relying on another intermediary. I haven’t stopped relying on APIs, but rather I’ve stopped relying on somebody else’s visual workflow builder to orchestrate them. AI has increasingly become that orchestration layer.

If my own buying behaviour has changed this much in such a short period, I doubt I’m alone. That’s why I think software companies are beginning to face a different challenge.

Features such as Zapier integrations, large integration marketplaces, and hundreds of pre-built connectors are still valuable, particularly for organisations that want dependable no-code automation managed by business users rather than developers. However, I suspect they’re becoming less influential than they once were. Increasingly, I find myself valuing well-designed APIs, full documentation, robust webhook support, SDKs, and architectures that are straightforward for both developers and AI to work with.

Ironically, the rise of AI may make APIs one of the most valuable product features a software company can invest in.

Before I finish, one point is worth emphasising. None of this has made me less enthusiastic about SaaS. If anything, it’s made me appreciate truly exceptional software even more. Platforms such as Pipedrive, Xero, Google Workspace, Adobe Creative Cloud, Brevo, Figma, Notion, and many others represent years, sometimes decades, of product development. They aren’t simply collections of features. They’re mature ecosystems, refined through millions of users, backed by dedicated engineering teams, extensive security, ongoing support, and continuous innovation.

Could AI help me recreate elements of those products? Absolutely. Could I realistically build something that actually competes with them in a reasonable amount of time? Almost certainly not.

In fact, AI has reinforced something I’ve always believed. The best SaaS products are valuable because they solve complex problems exceptionally well. They save time, reduce risk, and allow companies to focus on what they do best. That’s exactly why Slick Media exists. We don’t recommend software simply because it exists; we partner with platforms that consistently demonstrate real value.

What AI has changed is the threshold. Software now has to deliver enough value that it’s clearly better than what can be built with AI in a few hours or days. I see that as a positive development. The very best software becomes even more valuable, while products that exist simply because they fill a small functional gap may find themselves under increasing pressure.

The Prompt Has Become the New Search Box

I don’t think integration platforms disappear.

In fact, I expect they’ll evolve rapidly, incorporating AI into almost every aspect of how workflows are created and managed. Companies like Zapier are already investing heavily in AI, and I’d be surprised if they weren’t among the innovators forming the next generation of automation.

What is changing is something more fundamental — the starting point.

Five years ago, my instinct was to search for software.

Today, my instinct is to describe the problem.

• Sometimes the answer is still Zapier.
• Sometimes it’s Make.
• Sometimes it’s a dedicated SaaS platform.

Increasingly, though, it’s a bespoke integration built around a well-designed API, with AI doing much of the heavy lifting.

The tool has become secondary. The outcome has become the priority.

Perhaps that’s the biggest lesson I’ve taken from AI-assisted development.

For years, software categories expanded because every new problem seemed to justify another product and another monthly subscription. Integration platforms proved essential because they filled the gap between applications that couldn’t easily communicate with one another.

Today, AI is beginning to narrow those gaps, not by replacing software, but by making it dramatically easier to connect, customise, and extend the software we already use, and that has fundamentally changed the way I evaluate technology. When I review a new platform today, I’m no longer asking only whether it has the features I need. I’m asking whether it has the openness, flexibility, and API-first design that will allow AI to build around it tomorrow.

For software companies, I believe that’s becoming an increasingly important distinction. For buyers, it means the conversation is changing too. The question is no longer simply:

“Which software should I buy?”

It’s increasingly becoming:

“Should I buy software, customise existing software, or build exactly what I need?”

I don’t think this is the end of Zapier, Make, or integration platforms, far from it. They solved a genuine problem, and they’ll almost certainly continue to progress alongside AI.

What has changed is me. A few years ago, my first instinct was to search for software. Today, my first instinct is to describe the problem. Sometimes the answer is still an integration platform. Increasingly, though, it’s an API, AI-assisted development, and a solution built specifically for what I’m trying to achieve. That’s why I believe we’re witnessing a fundamental shift, not in automation itself, but in the way we think about creating it.

For years, Google was where I searched for software solutions. Today, I increasingly start somewhere else. I open an AI chat and describe the problem.

The prompt hasn’t just become the new search box. For many of us, it’s becoming the new development environment.