SaaS or custom build: what survives the AI wave?

SaaS only survives the AI wave if it is truly irreplaceable

Many Dutch business owners pay monthly for ten, fifteen, sometimes twenty different SaaS tools. One for email, one for project management, one for invoicing, one for customer follow-up. Every tool once did something specific that you could not build yourself. That logic is becoming less valid by the month.

AI dramatically lowers the barrier to building custom software. What once required a team of developers and months of work can now be delivered in weeks. That has a direct impact on how you should think about your current toolkit.

What the so-called SaaS-pocalypse actually means

The term has been circulating in tech circles for a while: SaaS tools that handle generic tasks are being displaced by AI agents and automations that do the same thing but with more flexibility, lower cost, and better integration into your own processes.

This does not apply to all SaaS. It applies to a specific category: tools you use for tasks that are repetitive, rule-driven, and data-dependent. Examples include:

• Manual lead follow-up in a CRM you maintain yourself
• Exporting reports from three separate systems and merging them in Excel
• Manually reminding clients about outstanding invoices
• Processing form submissions by hand into another tool

These are exactly the processes where automation hits hardest. Not because the SaaS tool is bad, but because the task itself disappears once you set things up properly.

When SaaS is still the right choice

Not every tool is vulnerable. SaaS survives where it offers real depth that is difficult to replicate. Examples include:

• Compliance and certification: tools that enforce GDPR processing, audit trails, or industry-specific standards
• Ecosystem and integrations: platforms with a broad network of connections built up over years
• Infrastructure scalability: hosting, security, and uptime you do not want to manage yourself
• Specialist functionality: CAD software, ERP for manufacturing companies, legal platforms

A simple test: does everyone in your industry use it, and would it take more than a year to build yourself? If yes, SaaS is probably still the right call.

When custom builds or automation make more sense

The balance shifts when you are buying SaaS tools for tasks that are really just business logic wrapped in a subscription. You pay for flexibility you never use, and you miss the integration you actually need.

In practice, the pattern looks like this:

1. You buy a tool to solve a specific problem
2. You adapt your workflow to the tool, instead of the other way around
3. You buy a second tool to fill the gap left by the first
4. You now manage five separate tools that do not communicate well

At that point the question is not whether to switch, but how quickly. An automation that connects your own data streams via n8n, an AI agent that handles customer communication, or custom software built around your exact workflow: all of these give you back control.

The difference with the standard approach is that at NRL Automations we do not start with the technology. We start with the business. First we look at where time and revenue are leaking, then we determine the smartest solution. Sometimes that is still an existing tool, but more often it is a combination of automation and custom work that fits better.

What AI agents concretely change in this picture

AI agents are software components that execute tasks autonomously, make decisions based on context, and can switch between systems. They communicate via APIs with your existing tools and can be equipped with access to your own data through RAG, a technique where the agent retrieves relevant information from a knowledge base before responding or taking action.

In practice: an agent that reviews incoming quote requests, qualifies them, sends an initial response, and creates the lead in your CRM, without any manual input. Or an agent that handles customer communication via WhatsApp or Instagram, trained on your products and policies, as described in the article on AI agents working with your own data.

These tasks used to require a combination of people and multiple SaaS tools. Now they can be handled with a well-configured automation.

The practical question: what do you do now?

Start with an audit of your current toolstack. Ask three questions per tool:

• What problem does this solve?
• Is that task distinctive to my business, or is it just administration?
• Could I handle this better with an agent or automated workflow?

Tools for unique, complex, or regulated tasks stay. Tools for generic tasks that dictate your workflow instead of supporting it deserve a second look.

The transition takes planning. You do not want to overhaul systems in the middle of a busy period. But waiting also has a cost: every month you pay for tools that your competitors have already replaced with faster, cheaper alternatives.

Whatever you choose, make sure everything is set up securely. Connections between systems, access rights, and data storage need to be configured correctly from day one. See how NRL Automations approaches this before anything goes live.

The foundation under SaaS is shifting

SaaS is not disappearing. But the part of the market built on simple task automation and loose integrations is under pressure. Businesses that now think carefully about which tools are truly irreplaceable and which can be replaced by automation will be in a stronger position. Not because they have the newest technology, but because their processes are sound.

Want to know which tools in your business could be the first candidates for automation? Plan a conversation and we will look at it together.