News & views
on AI

France just committed €655 million to sovereign AI. The headline is the money, but the real story is the example it sets: when Europe's largest companies and governments choose European models, infrastructure, and software, the whole ecosystem compounds.

Everyone talks about AI, but how do you get real and impactful solutions from the drawing board to fundamentally improve the way people work? Wolters Kluwer Schulinck and Mozaik have been partnering from ideation to commercial launch of highly successful AI applications. But let's be honest, it was no walk in the park.

We believe that AI needs to be in the form of end-to-end applications that are fully integrated in working processes and a joy to use. That's why we're excited to partner with SST Software.

In the rapidly evolving field of AI engineering, we're constantly seeking ways to move beyond simple code completion and leverage Large Language Models (LLMs) for more complex, project-specific development tasks.

In the current landscape of Applied AI engineering, ensuring consistent, reliable output from LLMs is a critical challenge. In several of our projects, we have had to balance accurate structured outputs with the inherent variability of LLMs. To solve this issue, we have developed a robust solution that leverages structured output and samples the most reliable response from multiple candidates.

We all have that drawer full of chargers at home. It took USB-C a decade to clean that up. Typically, vendors only embrace interoperability when market forces demand it. But AI flips that script.

AI agents are doing more than just talking; they're executing decisions that keep businesses nimble. From optimizing inventory to automating customer follow-ups, these tools are reshaping task management. Dive into how AI agents streamline operations while keeping your data secure.

AI is a powerful opportunity for content and software providers. But risks are emerging for the ownership of the customer interface, creation of new potential competitors and control of data. How can content & software providers capture the AI potential, without the downside?

Over a decade has passed since the ‘deep learning revolution’ turned the world of computer vision upside-down, and now we are about to witness another transformative shift that promises to make computer vision simpler and more accessible than ever before.

We have developed data platforms and products with perception that the final end-user is someone sitting behind a computer screen, clicking around in data tools and graphical presentations of insights. This is all about to change.

In the landscape of artificial intelligence, Large Language Models (LLMs) are rapidly changing how businesses operate and engage with data. It’s not just about enhancing productivity but also opening new avenues for innovation. We dive into the five use cases of LLMs, showcasing how they work and how they impact businesses.

Today, we’re very excited to introduce Mozaik to define, design, and develop groundbreaking AI. This is the start of the most exciting time in data technology, we help organizations navigate this from start to finish.

We’re excited to introduce Stijn Tonk as partner at Mozaik, joining Gert Jan Spriensma and Vincent Hoogsteder on our management team. Stijn brings extensive experience in building scalable platforms, leading teams, enhancing data & AI capabilities for some of world’s largest organizations.

Many people know about major players like OpenAI, which release cutting-edge models, but running these models often comes with high costs. Meanwhile, there's an increasing variety of other models emerging, leading to confusion in navigating the expanding landscape.

In the past months, we helped one of our customers bring a Retrieval-Augmented Generation (RAG) prototype into production. In this post, I would like to share our key learnings and milestones from this two-month adventure.

The first AI project in any company is the chance to deliver on high expectations and get folks excited for more. We believe you can do this by preventing an isolated case, and build a Cornerstone use-case instead.

This is a rare moment, where companies that have waited to make their move to a data warehouse have an advantage in their AI development. Why? Let’s dig into what technology works best for AI.

The way we search the internet is changing rapidly. AI tools like Perplexity, ChatGPT, Gemini and Claude are transforming how we find and interact with information online.

We've entered the era of choice in AI. After a year of OpenAI dominance, you now have the freedom to use equally performing models, at the same price range, from vendors ranging from Silicon Valley to Paris. It's now possible to build use-cases on hybrid models and overall pricing has been dropping 4x in just 15 months. It's a great time to build in AI.

When meeting with clients who have been briefed internally, we frequently observe a basic grasp of terms like vector embeddings. However, this understanding typically focuses on why it's necessary, rather than on its operational mechanics or the requirements for its implementation.

If you've been exploring AI, particularly Large Language Models (LLMs), you've likely encountered the term "token." This concept often comes up in discussions about pricing, where fees are typically based on a per-thousand-tokens model.