For the past couple of decades, software decisions in larger companies have followed a fairly straightforward logic. Standard needs were covered with off-the-shelf subscription software, or SaaS, while custom development was considered only when the software supported a truly distinctive business process, created a competitive advantage, or when no suitable solution existed on the market, says Raimo Seero, CTO of Uptime.
Artificial intelligence is now changing that logic. As the total cost of developing custom solutions has fallen, companies need to look beyond the initial price tag. Increasingly, they must also assess long-term dependence on a vendor, regulatory requirements, the cost of future changes, and how quickly the software can adapt to the company’s evolving needs. In other words, we are reaching a point where SaaS is no longer the automatic default choice, but rather a conscious trade-off between different risks and benefits.
This does not mean that off-the-shelf software has had its day. On the contrary, in many cases it remains the most sensible option. However, AI has made the development of custom software significantly faster and more cost-effective, which means it is now something companies should seriously consider.
This shift is already visible in practice. According to a study by the software development platform Retool, more than one-third of companies have replaced at least one SaaS solution with software they developed themselves, and 78% plan to build more applications in-house in the near future. Most often, this concerns internal tools, workflow automation, and the integration of different systems.
The Total Cost of SaaS Goes Beyond the Licence Fee
The biggest advantage of off-the-shelf software remains speed of implementation. At the same time, costs and risks that only become apparent over the years are often underestimated at the time of purchase.
Over a five- to eight-year period, licence fees can become a significant expense, especially in organisations with hundreds of users. In addition, pricing is no longer based solely on the number of users. Increasingly, companies face paid AI features, new pricing packages, and price changes over which the customer has little or no control.
Vendor dependency is just as important. The longer a company uses a single platform, the more difficult it becomes to move away from it. Data, integrations, and workflows become tied to that particular software, and switching to a new solution often means a costly migration project. At the same time, the software provider may change its pricing, discontinue the development of certain features, or shift its product priorities.
Regulatory requirements are also becoming increasingly important. In the European Union, new legislation such as NIS2, the Data Act, and the AI Act is increasing companies’ responsibility for data processing and information security. It is no longer enough to know that data is stored on European servers. It is equally important to understand which jurisdiction the service provider falls under and who may have the right to access the company’s data.
AI Is Making Custom Solutions Increasingly Viable
In the past, a large share of software development was spent on building technical foundations and carrying out routine work. Today, more time and budget can be directed towards solving the processes that actually create value for the business. This means that the same investment can lead to a working solution faster, and even smaller projects or those aimed at a narrower business need can now pay off.
The strongest argument for a custom solution emerges when the software is strategic for the company. The more important a system is to core operations, competitive advantage, or company-specific workflows, the more the benefits of a custom solution outweigh the convenience of off-the-shelf software. In such cases, flexibility, control over development priorities, and the ability to make changes when the business needs them – rather than when the software vendor releases them in the next version – become decisive.
This does not mean that every system should be built in-house. If the need is standard, the solution changes rarely, and it is not directly linked to the company’s competitive advantage, SaaS often remains a sensible choice. Even then, however, it is worth doing the calculation. If the software is highly standardised but its licence costs become substantial over the years, replacing it with a simpler custom-built solution may make economic sense even if the functionality does not change significantly.
A Real Example: Why We Built Our Own Time Management Software
At Uptime, we used licence-based time management software for around 150 employees, but in our day-to-day work we only needed a small portion of its functionality. The rest added cost and complexity for the company without creating value.
With the help of AI, we mapped the actual requirements for time tracking and developed a new system that includes only the features employees use in their daily work. We integrated the solution with our existing workflows, and its architecture allows us to add new features as needed, without having to wait for the software vendor’s next version updates.
The investment paid for itself in approximately two years. At the same time, we reduced our dependence on an external software provider and gained greater control over both the development of the system and its future costs. Thanks to AI and modern development methods, making smaller changes has also become significantly faster and more cost-effective than before.
