Smart Manufacturing & Materials-2.jpg

<p><strong>More and more startups are developing products based on large language models (LLMs) and generative AI. While a working prototype can be built quickly these days, a scalable and reliable AI product requires much more than just a clever prompt. That’s why Datakami, based in Nijmegen, helps AI startups not only with technical challenges but, above all, with building a solid foundation. To make that knowledge more widely available, the company developed the AI Readiness Score: a free self-assessment that allows entrepreneurs to evaluate how future-proof their AI product is.</strong></p><p><img src="https://cms.aihub-oost.nl/uploads/judith_van_stegeren_datakami_landscape_397f57d74b.jpg?token=kCwGqVh-mHXusEW9E207T9CcnG8jhpUbQ8cH0Vs9pEM&amp;expires=1787832286" alt="judith-van-stegeren-datakami-landscape.jpg" srcset="https://cms.aihub-oost.nl/uploads/thumbnail_judith_van_stegeren_datakami_landscape_397f57d74b.jpg?token=qgDW-um3M97Y-Gv045CMNshNXtgW1H8uAWh1aE7muLs&amp;expires=1787832286 209w,https://cms.aihub-oost.nl/uploads/small_judith_van_stegeren_datakami_landscape_397f57d74b.jpg?token=5y0vPS10QAFq4yXmp9lxKEuW5ARGLmrgU5LcCMZ1ce0&amp;expires=1787832286 500w,https://cms.aihub-oost.nl/uploads/medium_judith_van_stegeren_datakami_landscape_397f57d74b.jpg?token=VwNrOV_BYLLW2y8ilMDcYhGIYIWTliHOd2Z6PgVJUKI&amp;expires=1787832286 750w,https://cms.aihub-oost.nl/uploads/large_judith_van_stegeren_datakami_landscape_397f57d74b.jpg?token=qvu9lS4Dtp6uVKySQjmJqPlvyl2zZVohcA1sY0D5REo&amp;expires=1787832286 1000w," sizes="100vw" width="2000" height="1499"></p><p>Early on in the rise of generative AI, Judith van Stegeren, founder of Datakami, recognized that startups were in need of specialized knowledge about AI engineering. From Nijmegen, she built Datakami into a company that focuses entirely on startups developing products on top of LLMs and other generative AI models. Datakami now works with AI startups across Europe and the United States, from seed funding through Series B.</p><h2>First the foundation, then optimization </h2><p>Datakami not only advises but also actively helps build AI products. “Many startups are actually building based on intuition. They don’t yet have a clear enough understanding of exactly what their AI models do and have a hard time determining whether a change to a prompt, model, or piece of code is actually an improvement,” says van Stegeren.</p><p>The issues startups bring to the table vary widely. Sometimes an AI model gives unexpected answers; other times, the costs of language models escalate rapidly, or the response time is too long. According to Van Stegeren, these are often symptoms of the same problem: the technical foundation isn’t mature enough yet.</p><p>That’s why Datakami usually doesn’t start by solving the problem itself, but by setting up a solid infrastructure. The team provides insight into the behavior of the AI models, measures the quality of the system, and configures the product to systematically collect user feedback. Only then can entrepreneurs make informed decisions about new models or improvements to their product.</p><h2>AI Readiness Score </h2><p>Because Datakami kept encountering the same challenges over and over again, the idea for the AI Readiness Score was born. The online assessment consists of a short list of yes/no questions and was developed specifically for startups building products with LLMs or similar generative AI models.</p><p>The questions cover topics such as observability, evaluation of AI output, user feedback, and reliance on external AI providers. Within minutes, entrepreneurs gain insight into the maturity of their AI product. They can also compare their score with other startups in the same growth stage.</p><p>“The AI Readiness Score is essentially a summary of everything we’ve learned over the past few years working with early-stage AI startups. I wanted to make that knowledge accessible to founders so they can check for themselves whether they’ve considered the most important engineering best practices.”</p><p>The AI Readiness Score is also a useful tool for investors, for example, as a starting point for technical due diligence. The questionnaire is short, but the answers quickly provide insight into a startup’s technical maturity and whether the team has a clear vision for addressing any outstanding issues.  </p><h2>Less Dependence on a Single AI Provider </h2><p>According to Van Stegeren, a solid technical foundation will only become more important in the coming years. Major AI labs are rapidly evolving their models, changing prices, and increasingly bringing new applications to market themselves. Startups that are entirely dependent on a single provider therefore run the risk of getting stuck. “As a startup, you don’t want to become a pawn of the big AI labs.”</p><p>By systematically collecting data from the start on how the product is used, how the models perform, and how effective the AI output is, startups can objectively compare different models and—if necessary—switch to another AI provider with relative ease. This helps them avoid vendor lock-in and maintain the flexibility to adapt to market developments.  </p><p>Such a solid foundation also gives startups a clear path forward: greater insight into their own product, better justification for technical choices, more freedom of choice, and greater agility. This allows founders to better protect their intellectual property and margins. It also makes a startup more attractive to investors.</p><h2>Tips for entrepreneurs developing AI products </h2><ul><li data-list-item-id="e9be4d21ddd7e4e12080124cff7818c41">First, lay a solid foundation. A prototype is just the beginning; make sure you understand exactly what your product and models do before you start expanding or optimizing.</li><li data-list-item-id="ee5e7ec2d1afc2ea780c9fea146e29bcd">Make improvements measurable. Set up evaluations that allow you to objectively determine whether a change is truly better for your users.</li><li data-list-item-id="eeb453afc2dbce9f26a4c07e1d9986d9f">Gather user feedback. This information is essential for making targeted improvements to AI products.</li><li data-list-item-id="e79504d4ee7dbbcfec383210754b141cd">Avoid vendor lock-in. Set up your infrastructure so that you can switch to another AI provider if necessary.</li><li data-list-item-id="eb0412b7367b857bf8322a75f3ca1f580">Remain critical of external dependencies. Regularly review which models and platforms you’re using and how much freedom of choice you still have.</li></ul><p>Datakami shows that successful AI products aren’t just about the latest models, but above all about a strong technical foundation. Those who invest in this from the start build an AI product that’s ready for the next phase of growth and for the rapid developments in the AI market.</p><h2>Curious about how future-proof your AI product is? </h2><p>Are you working on an AI product that uses LLMs or other generative AI models? With the AI Readiness Score, you can discover in about five minutes where your technical foundation is already strong and where there are still opportunities for improvement. Afterward, you’ll receive a personalized gap analysis with concrete next steps. You can take the assessment for free at: https://www.datakami.com/ai-readiness-score. </p>

Datakami helps AI startups build future-proof AI products

These entrepreneurs have paved the way for you

Meet our frontrunners

judith-van-stegeren-datakami-landscape.jpg

Datakami helps AI startups build a future-proof AI product

<p>More and more startups are developing products based on large language models (LLMs) and generative AI. While a working prototype can be built quickly these days, a scalable and reliable AI product requires much more than just a clever prompt. That’s why Datakami, based in Nijmegen, helps AI startups not only with technical challenges but, above all, with building a solid foundation. To make this knowledge more widely available, the company developed the AI Readiness Score: a free self-assessment tool that allows entrepreneurs to evaluate how future-proof their AI product is.&nbsp;</p>