Custom software or SaaS? What AI changed in this decision

Nov 2026 · 7 min read

The question "will AI kill SaaS?" took over the market in 2026. For those who invest in technology, the useful question is a different one: does the off-the-shelf software the company pays for today solve its operational pain points? With AI making building cheaper, custom software is once again a real option, as long as it starts from the right problem.

Summary

  • The debate about the end of SaaS points to a market correction, not to its end, and reopens the conversation about custom software.
  • Some symptoms show when software does not fit the operation: one subscription for every problem, idle licenses, processes bent to the tool and price increases with no gain.
  • AI has lowered the cost of building, but security, integration and ongoing support still require engineering.
  • The classic criteria still apply, but AI has changed the weight on each side, and niche tools are the first candidates to become custom software.

AI will not kill SaaS, but it has changed the math for software buyers

According to Tech Brew, in 2026 investors began selling shares of SaaS companies (software as a service, the per-user subscription model) from the end of January. The bet is that AI agents and coding tools will make that model obsolete.

The article itself downplays the panic. The people interviewed talk more about a correction than a collapse. Large companies such as Walmart and LiveRamp are building more internal tools, and faster, but have not yet canceled their SaaS contracts.

The point that interests us most is a different one. Bill Vass, CTO of Booz Allen, recalls that before ERPs became standard in the 1990s, companies built their own software. In his view, AI may take the market back to that kind of customization.

So the question for leadership is not whether SaaS will end. It is whether the software the company pays for today speaks to its pain points.

Symptoms that your software does not fit your operation

Not every annoyance with a system justifies switching tools. But some symptoms, when they show up together, mean the company is paying for software that does not keep up with the way it works.

  1. One subscription for every small problem. Each area signed up for its own niche tool, and there is little visibility into how many there are, how much they cost together and how much they overlap.
  2. Paid licenses that go unused. The company pays per user, but some people use a fraction of the features, or only log in to copy information somewhere else.
  3. The process has bent to the tool. Steps that would make sense for the business were cut or worked around because the system does not allow them.
  4. Improvements depend on the vendor’s queue. The change the operation needs becomes a support ticket and waits for the vendor to decide whether it fits its priorities.
  5. Price increases with no gain. The contract gets more expensive at renewal, but the operational problem stays the same.

The last point has a market figure behind it. According to Gartner data cited by Tech Brew, software contract renewals have come with increases of 10% to 20%. When cost goes up and the pain stays, it is time to redo the math.

Why this decision is different with AI

Building software has become faster. The same Tech Brew article reports that two CNBC reporters with no coding experience built a copy of a Monday.com dashboard in less than an hour. The interface layer, the screens and dashboards, is getting easier and easier to reproduce.

This has two effects for decision makers. First, the cost of building something custom has dropped, and so has the cost of switching vendors. Second, a screen built in an hour is not a system ready for the operation.

Alexey Korotich, chief product officer at Wrike, sums up the difference in the article: the interface is easy to copy, but the infrastructure and reliability are not. Security, data governance, integration with existing systems and support after launch still require engineering.

We have covered this on the blog in I built an MVP with AI, now what?: getting it live and keeping it running is where most of the work shows up.

What AI changed in the choice between SaaS and custom software

The classic criteria still apply. Market-standard processes tend to stay on off-the-shelf tools, and processes that set the company apart tend to call for their own system. We go deeper into this choice in Custom admin panel or dashboard: when it is worth building.

What AI changed is the weight on each side of the scale. With building now cheaper, custom software competes in cases that used to be out of the question. And SaaS comes under more pressure exactly where it is weakest: the Tech Brew article indicates that point and specialized solutions feel the pressure more than large, integrated platforms.

In practice, the first candidate to become custom software is usually that niche tool that solves half the problem and gets more expensive at every renewal. Broad platforms that support standard processes tend to stay where they are.

How Espresso Labs helps you decide on and build custom software

If you recognized two or more symptoms, the next step is not to start building. It is to understand which pain is worth tackling first and the simplest way to solve it, which is sometimes adjusting the SaaS you already have. That is why we start with Discovery & Design, before any line of code.

What we bring to this decision:

  • Discovery to map the pain points: we understand the process, the tools in use and where the rework is, to prioritize what really needs its own system.
  • From discovery to delivery: development builds on what was mapped, without losing context between phases.
  • Integration with what already works: custom software can coexist with the SaaS tools that still make sense, instead of replacing everything.
  • Support after launch: maintenance and hosting so the system stays secure and keeps evolving.

One example from our portfolio is the Acer case: a partner portal with a catalog and admin panel that gave the internal team the autonomy to update content without depending on developers. On the support side, we look after more than 57 systems under continuous maintenance.

Want to find out whether your operation calls for custom software? Talk to a specialist through our contact form or on WhatsApp.

Frequently asked questions

What is custom software?

It is a system developed for a specific company, based on its processes, rules and integrations. Instead of the team adapting to an off-the-shelf tool, the software is designed around the operation. The company also gains the freedom to evolve the system at the pace of the business, without depending on a SaaS vendor’s priorities.

Will AI kill SaaS?

Everything suggests it will not. The 2026 debate points more to a market correction, with pressure on prices and on niche tools, than to the end of the model. What changes is the math: with AI, building has become cheaper, and custom software now competes with SaaS in more situations than before.

Is custom software more expensive than SaaS?

It depends on the time horizon and the problem. SaaS has a lower entry cost, but it adds up licenses, price increases and the rework of people working around the tool. Custom software requires an upfront investment and ongoing support, but it solves the specific pain. A fair comparison looks at these costs over a few years, not just the first one.

Can you use custom software and SaaS at the same time?

Yes, and that is the most common scenario. Standard processes, such as email and accounting, usually stay on SaaS. Custom software comes in where the differentiator is, or where there is a pain no off-the-shelf tool solves, integrated with the systems the company already uses. That way, the investment goes where the return tends to be highest.

Where should I start if I suspect our software is not working for us?

Start by mapping where the team loses time: overlapping tools, idle licenses, steps worked around, changes stuck in the vendor’s queue. With that map, it is easier to decide whether to adjust the current tool, integrate systems or build something of your own. That is the role of discovery: understanding the problem before investing in the solution.

Does your operation call for custom software?

Tell us which tools and processes are holding your operation back. We will have a quick conversation to understand the scope and present a path with an estimate, no strings attached.