Go to content

Software and technology trends for 2025

Blog

Blog

Date  16 January 2025
5 min read
by Ron Rademaker

Software and technology trends for 2025

What will 2025 bring for us and our clients? In this blog, we’ll take you through some of the trends we foresee, what these trends mean for our clients, and how Harborn can help you leverage them.

Generative AI

A tech trends blog about 2025 wouldn’t be complete without a section on generative AI, also known as Large Language Models (LLMs). So, let’s dive right in. After ChatGPT broke the record of 100 million active users just two months after its launch in early 2023, the use of AI in daily tasks became normalized in 2024. But can AI live up to the sky-high expectations? And what will 2025 bring?

The AI Act

Starting in 2025, new legislation will regulate the use of AI. Some AI applications will be prohibited under these rules. At the moment, we don’t see any direct impact on our clients’ use cases. However, the high-risk category is relevant for many of our clients. For example, new rules will apply to matching candidates with job vacancies. Harborn can assist in meeting the required regulations to ensure such applications can be used legally from a technical perspective.

Through our innovation program, Harbase, we are developing tools to make AI usage more transparent, simplifying human oversight of AI systems. Such oversight will become mandatory for all high-risk AI applications starting in August 2025.

Transactional AI (AI Agents)

When we think of generative AI today, we often think of ChatGPT and other chatbots, typically with added functionalities like internet search or the ability to use proprietary data sources for more targeted applications. In 2025, we expect AI functionalities to expand further, including the ability to handle transactions. Initially, these will likely involve low-risk transactions or those requiring human oversight, such as scheduling appointments or making restaurant reservations. As trust in AI grows and error rates decrease, more impactful transactions will also become part of AI’s capabilities.

Media Creation

Generative AI is much more than just ChatGPT. Applications like Midjourney, Sora, and DALL-E are increasingly capable of creating realistic images and videos. This allows clients to produce media faster and at lower costs, perfectly aligned with their goals. When specific elements, such as certain individuals or environments, are not required, generative AI offers a highly realistic alternative to traditional shoots, eliminating the associated costs, hassle, and logistics. Harborn helps clients create visual content tailored to their needs, and once Sora becomes available in the Netherlands, we’ll expand our services to include video production.

Domain-Specific Models

The rise of LLMs has led to a surge in demand for specialized processors (GPUs), primarily produced by NVIDIA. This has caused shortages in the market and driven NVIDIA’s stock price up tenfold since early 2023. As models grow larger, they require significant time, energy, and processing power to train. To address this, we expect to see more domain-specific models emerge, enabling faster, cheaper development with less reliance on suppliers like NVIDIA. For example:

  • TAT-LLM 7B specializes in answering financial data questions.
  • ALMA 13B focuses on translations.
  • Code Llama 13B Python is designed for writing Python code.

These models are small enough to run on a powerful laptop, allowing clients to choose a model that better fits their use case. This results in comparable or better outcomes at significantly lower operational costs. Through our Harbase innovation program, Harborn is developing modules that make it easy to switch between models, so clients don’t have to wait for a domain-specific model to become available. Additionally, this tooling provides direct insights into the costs of an LLM, making the financial impact measurable and transparent.

Multi-Cloud

With AI trends outlined, let’s shift our focus to the next horizon: how multi-cloud environments make organizations more flexible and future-proof. For years, Amazon Web Services (AWS) has been the market leader in cloud services. However, in recent years, Azure has caught up completely. It’s becoming increasingly important to run business-critical applications in multi-cloud environments to reduce costs, minimize dependency on a single provider, and ensure the availability of critical applications.

Cost Optimization

With the rise of AI models, which often run in and are accessible via the cloud, cloud expenses are expected to accelerate in the coming years. Our clients are also seeing an increase in their costs. Harborn helps by providing advice, hands-on optimizations in cloud environments, and assistance with cloud migrations. Our accessible quick scan quickly identifies realistic cost-saving opportunities.

Vendor Lock-In

Cloud services can generally be divided into two categories: general services and cloud-specific services. While cloud-specific services can save time and effort, they also create dependency on a single provider. Harborn prevents this dependency by using cloud-agnostic tools on top of the cloud, without sacrificing the benefits of cloud-specific services.

Increased Availability

More organizations are running business-critical applications in the cloud. As a result, the risks of cloud unavailability are becoming increasingly unacceptable, making dependency on a single provider undesirable. A multi-cloud solution can mitigate this risk, provided there’s no reliance on a specific cloud service. Harborn offers a cloud-agnostic platform that makes applications compatible with any cloud. All our applications are designed to work in any cloud or multi-cloud environment, even if the need for a multi-cloud solution isn’t yet pressing.

Digital Twin Technology

A digital twin is a virtual replica of physical objects, processes, or systems. This technology enables process optimization, predictive maintenance, and risk-free simulations. In 2025, the use of digital twins will continue to grow, particularly in industries like manufacturing, healthcare, and infrastructure. Harborn helps clients integrate digital twins with real-time data, allowing them to manage physical processes more efficiently and simulate future scenarios.

Phygital Experiences

The line between the physical and digital worlds is becoming increasingly blurred. Phygital experiences—a blend of physical and digital interactions—offer new ways to engage customers and surprise users. Examples include:

  • Augmented reality (AR) for virtual product testing.
  • Virtual reality (VR) for immersive training.
  • Extended reality (XR) for interactive brand experiences.

In 2025, we expect further growth in the use of these technologies, particularly in retail, events, and education. Harborn supports clients in developing innovative phygital solutions that deliver seamless and impactful user experiences.

The future

The technological outlook for 2025 presents organizations with unprecedented opportunities to grow faster and strengthen their competitive position. Generative AI and transactional AI not only enable more efficient work and creative content use but also build trust in automated processes. Domain-specific models allow for tailored solutions, while multi-cloud strategies reduce dependency on major providers and mitigate risks for organizations.

“At Harborn, we believe real change starts with addressing the right challenges. Our teams work as a ‘special force’ in experience, growth, and engineering to solve complex problems and achieve sustainable growth.”

Want to know how we can help your organization with digital transformation, customer engagement, or growth marketing?

Feel free to contact us—we’d love to discuss the possibilities!

+31 10 436 50 50