2025技术雷达-针对当今科技领域发展的前沿指南（第32期）（英）

An opinionated guide totoday’s technology landscape About the RadarRadar at a glanceContributorsProduction CreditsThemesThe Radar About the Radar Thoughtworkers are passionate about technology. We build it, research it, test it, open source it,write about it and constantly aim to improve it — for everyone. Our mission is to champion softwareexcellence and revolutionize IT. We create and share the Thoughtworks Technology Radar in supportof that mission. The Thoughtworks Technology Advisory Board, a group of senior technology leaders The Radar captures the output of the Technology Advisory Board’s discussions in a format thatprovides value to a wide range of stakeholders, from developers to CTOs. The content is intended We encourage you to explore these technologies. The Radar is graphical in nature, grouping itemsinto techniques, tools, platforms and languages and frameworks. When Radar items could appearin multiple quadrants, we chose the one that seemed most appropriate. We further group these Radar at a glance The Radar is all about tracking interesting things, which we refer to as blips. We organize the blips inthe Radar using two categorizing elements: quadrants and rings. The quadrants represent different A blip is a technology or technique that plays a role in software development. Blips are “in motion”— their position in the Radar often changes — usually indicating our increasing confidence in Adopt:We feel strongly that the industryshould be adopting these items. We use Trial:Worth pursuing. It’s important tounderstand how to build up this capability.Enterprises can try this technology on a Assess:Worth exploring with the goal ofunderstanding how it will affect Hold:Proceed with caution. NewMoved in/out Our Radar is forward-looking. To make room for new items, we fade items that haven’t movedrecently, which isn’t a reflection on their value but rather on our limited Radar real estate. Contributors The Technology Advisory Board (TAB) is a group of 21 senior technologists at Thoughtworks.The TAB meets twice a year face-to-face and biweekly virtually. Its primary role is to be an The TAB acts as a broad body that can look at topics that affect technology and technologistsat Thoughtworks. This edition of the Thoughtworks Technology Radar is based on a meeting CamillaFalconi Crispim NimishaAsthagiri Production Credits Design and Multimedia •Willian Amaral—Product Owner•Preeti Mishra—Project andCampaign Manager•Richard Gall—Content Editor •Leticia Nunes— Lead Designer•Sruba Deb— Visual Designer•Kevin Barry— Multimedia Specialist•Ryan Cambage— Multimedia Specialist Digital and Web Experience Communications •Rashmi Naganur— Business Analyst•Brigitte Britten-Kelly— DigitalContent Strategist•Vandita Kamboj— UX Designer•Anisha Thampy— Visual Designer •Shalini Jagadish— InternalCommunications Specialist•Hiral Shah— Social Media Specialist•Abhishek Kasegaonkar— SocialMedia Specialist Themes Supervised agents in coding assistants Two of our themes highlight the rapid innovation in generative AI, and one of them is about theaccelerating capabilities of coding assistants. More and more of these tools allow developers to driveimplementation directly from an AI chat within their IDE — a mode also called “agentic”, “prompt-to-code” or “chat-oriented programming (CHOP).” In this approach, AI assistants go beyond answeringquestions or generating small snippets; they navigate and modify code, update tests, executecommands and, in some cases, proactively fix linting and compilation errors. While we remain skepticalof coding agents that promise fully autonomous development of large tasks, we’ve seen promising Evolving observability We’ve seen significant movement in the observability space, driven by the growing complexityof distributed architectures. While observability has long been essential, it continues to evolvealongside the rest of the software development ecosystem. One emerging focus is LLM observability,a critical piece in operationalizing AI. We’ve seen a surge in tools for monitoring and evaluatingLLM performance, includingWeights & Biases Weave,Arize Phoenix,HeliconeandHumanLoop.Another trend is the integration of AI-assisted observability, where tools leverage AI to enhance R in RAG We expect different aspects of the generative AI ecosystem to evolve at varying rates, and in thisedition of the Radar, we see this happening for the R in RAG (retrieval-augmented generation). One ofthe key interactions with the LLM black box is customizing the prompt’s inputs to generate relevant dynamically adjusts responses based on feedback or heuristics; Fusion-RAG, which combines multiplesources and retrieval strategies for more comprehensive and robust responses; and Self-RAG, whichavoids the retrieval step altogether, fetching data on demand. We also highlighted FastGraphRAG, Taming the data frontier Big data has long been a key concern for the industry, but conversations around this ed