Abstract
Foundation models (FMs) such as large language models (LLMs) havesignificantly impacted many fields, including software engineering (SE). Theinteraction between SE and FMs has led to the integration of FMs into SEpractices (FM4SE) and the application of SE methodologies to FMs (SE4FM). Whileseveral literature surveys exist on academic contributions to these trends, weare the first to provide a practitioner's view. We analyze 155 FM4SE and 997SE4FM blog posts from leading technology companies, leveraging an FM-poweredsurveying approach to systematically label and summarize the discussedactivities and tasks. We observed that while code generation is the mostprominent FM4SE task, FMs are leveraged for many other SE activities such ascode understanding, summarization, and API recommendation. The majority of blogposts on SE4FM are about model deployment & operation, and system architecture& orchestration. Although the emphasis is on cloud deployments, there is agrowing interest in compressing FMs and deploying them on smaller devices suchas edge or mobile devices. We outline eight future research directions inspiredby our gained insights, aiming to bridge the gap between academic findings andreal-world applications. Our study not only enriches the body of knowledge onpractical applications of FM4SE and SE4FM but also demonstrates the utility ofFMs as a powerful and efficient approach in conducting literature surveyswithin technical and grey literature domains. Our dataset, results, code andused prompts can be found in our online replication package athttps://github.com/SAILResearch/fmse-blogs.