The 2024 Edition of Thoughtworks Technology Radar Is Out!

It is one of our favourite sources of trends and ideas and was an inspiration for our own ARTIN Tech Radar a while ago. And since there are several areas and practical tips, this edition is obviously dominated by AI and Large Language Models (LLMs).

Not really a surprise as LLMs seem to be the next big thing after sliced bread. Still, it is interesting how it appears in so many different contexts through all four quadrants of the radar. LLM usage has already moved beyond a simple question-answering bot. The presence of Retrieval-Augmented Generation (RAG) in the Adopt ring shows, that LLM works best when it's a component of a bigger tool, not the tool itself.

And the radar also suggests tools to help implement it, like Dify. Dify is an open-source LLM app development platform and it's intuitive interface combines AI workflow, RAG pipeline, agent capabilities, model management, observability features, and more, letting you quickly go from prototype to production.

The radar also suggests, that it is better to use AI as a team assistant, rather than a personal one. Providing it with the context of the team (such as knowledge sources for RAG) or even implementing Chat OPS, will improve the experience as such. Yet there are some pitfalls of LLMs that you should be aware of - at least according to the radar. Thoughtworks warns against a rush to fine-tune LLMs, that in most cases, can be better implemented using RAG. Even overenthusiastic LLM should be carefully avoided as sometimes there are better or more correct approaches to the problem at hand.

Cannot find LangChain?

The most surprising single point on the radar is moving the LangChain framework to the Hold ring. LangChain is a trendy tool among developers, however, according to the radar, it's getting overcomplicated and bloated. Radar recommends considering more direct and lightweight libraries, like LiteLLM. With all that said about LLMs, the radar is not only about it. You can notice hints about ARM in the cloud or a handy tip for a financial accounting database Tigerbeetle. The distributed financial transactions database designed for mission-critical safety and performance.

All of these tips and suggestions make the 2024 Technology Radar a rich and opinionated source for developers and all tech enthusiasts. It therefore becomes a practical guide to today’s technology landscape that should not be overlooked.

Did you get inspired?

Jaroslav Müller

Developer

Jaroslav Muller is a seasoned developer, technology enthusiast, and team leader currently working at ARTIN in Brno, Czechia. With over eight years at ARTIN, he has held roles including DevOps, architect, and team lead, focusing on bootstrapping project architectures, selecting appropriate tools, and ensuring project scalability. He also contributes to the company's technology strategy and recruitment efforts.
Jaroslav Müller

We're here as well