Blog

Intelligent assistants, automated decisions, content at the click of a button. AI has long been part of everyday life and is already shaping how we work, lead and think. Even so, many questions remain open: where does sensible use begin, and where does the hype end? And what attitude do companies need in order to realise the technology's potential responsibly?

Without any complex special solutions, you can build a high-performance vector database for AI applications with "conventional" PostgreSQL. Specialised vector indexes make sure that queries are answered extremely fast. That saves resources, keeps your AI systems and agents responsive, and minimises operational risk by relying on standard software.

Upgrading to Tailwind CSS v4 makes our development work more pleasant and gives us new features such as container queries along with a faster development environment. With the focus on CSS variables, the detour via Sass falls away and we use more native CSS functions instead.

Open source LLMs are developing rapidly and increasingly reaching a level of performance comparable to closed source models such as GPT-4. Large models like DeepSeek R1 (671B) deliver impressive results, but they require powerful hardware. Smaller models such as Phi-4 (14B) are easier on resources but come with some limitations in quality.

Manual tests are often time-consuming and error-prone. At SPRYLAB we use the Playwright code generator, a solution that lets us build end-to-end tests (E2E) in next to no time. That saves time and resources and delivers higher quality in our projects. In this blog post I would like to give you a brief look at how we use Playwright.