In a constantly evolving technological world, open source frameworks dedicated to artificial intelligence (AI) demonstrate innovation and adaptability. As agentic AI begins to take shape, various projects are emerging to strengthen the connectivity between agents and IT infrastructure. In 2025, the need for robust and interoperable solutions is more crucial than ever, with initiatives marked by community and strategic collaborations. This article examines recent advances in the sector, including the Kagent and Dapr projects, while exploring the impact of these developments on the resilience and efficiency of cloud-native systems.<\/p>\n\n
Launched in March 2025, Kagent positions itself as a catalyst for innovation for AI agents within cloud infrastructures. Developed by Solo.io<\/strong>, this framework aims to be a bridge between autonomous agents and complex IT environments. Lin Sun, Head of Open Source at Solo.io, emphasizes that the idea for Kagent emerged from needs identified during client engagements, where the need for expert assistance was felt. The goal is clear: to provide a form of \u00ab\u00a0clone\u00a0\u00bb of expert developers to relieve support teams without disrupting the development process.<\/p>\n\n Kagent combines several essential elements for seamless integration into cloud infrastructures. The key feature is its declarative API, which facilitates interaction between agents and various cloud resources. By offering pre-integrated plugins for tools like Kubernetes, Helm, Prometheus, and more, Kagent significantly reduces deployment complexity. Here are some key elements that define Kagent’s architecture:<\/strong>Declarative API: Intuitive interface allowing developers to easily define and manage agents.<\/strong>Built-in plugins: Immediate access to popular tools and services within cloud environments.<\/strong> Kubernetes-first design:<\/p>\n\n Leverages the flexibility and scalability of Kubernetes to run agents wherever the tool is deployed.<\/p>\n\n AutoGen<\/p>\n\n API Declarative<\/strong> Conversational Orchestration<\/strong> Integration<\/p>\n\n Consistent with Kubernetes<\/p>\n\n Integration plans for various SDKs, ranging from Python to Java. This flexibility allows components to be shared between different cloud service providers and databases, facilitating adoption and continuous improvement of the system.<\/p>\n\n The combination of Dapr with its new AI agent capabilities could represent a decisive advancement for microservices architecture. Indeed, through its ability to seamlessly orchestrate AI agent workflows, Dapr allows companies to leverage modern machine learning tools such as TensorFlow or PyTorch, thus integrating complex models into a managed ecosystem. This promises not only an improvement in the performance of AI systems but also the possibility of integrating advanced resilience features directly into the core of applications. To better understand this evolution, here is a comparative table illustrating the characteristics of Dapr Agents in relation to its previous versions:<\/strong>Characteristic<\/p>\n\n Dapr Agents<\/p>\n\n Service Orchestration<\/p>\n\n Adapted to AI agents This clearly shows how Dapr Agents is part of a desire for adaptability and efficiency, meeting the growing needs of modern infrastructures.<\/strong> Working with the Semantic Kernel and its developments Alongside the advances in Kagent and Dapr, Microsoft has integrated a promising technology into AutoGen:<\/strong>Semantic Kernel Process Framework (SKPF)<\/p>\n\n . Launched in AutoGen version 0.4 in January 2025, this new feature has the potential to expand the scope of interaction between AI agents and enterprise applications.<\/p>\n\n <\/p>\n\n <\/strong><\/p>\n\n <\/p>\n\n <\/p>\n\n <\/p>\n\n <\/p>\n\n\n","protected":false},"excerpt":{"rendered":" In a constantly evolving technological world, open source frameworks dedicated to artificial intelligence (AI) demonstrate innovation and adaptability. As agentic AI begins to take shape, various projects are emerging to strengthen the connectivity between agents and IT infrastructure. In 2025, the need for robust and interoperable solutions is more crucial than ever, with initiatives marked […]<\/p>\n","protected":false},"author":1,"featured_media":10108,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[41],"tags":[149,19318,19324,19321],"class_list":["post-10118","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-non-classe-en","tag-artificial-intelligence-en","tag-evolution-of-technologies-en","tag-it-infrastructure-en","tag-open-source-frameworks-en"],"_links":{"self":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts\/10118","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/comments?post=10118"}],"version-history":[{"count":1,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts\/10118\/revisions"}],"predecessor-version":[{"id":10119,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/posts\/10118\/revisions\/10119"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/media\/10108"}],"wp:attachment":[{"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/media?parent=10118"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/categories?post=10118"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.mon-agent-ia.fr\/blog\/wp-json\/wp\/v2\/tags?post=10118"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Kagent Features and Architecture<\/h3>\n\n
Dapr<\/h3>\n\n
\n\n
\n \nLimited to AutoGen<\/th>\n Adaptable to various environments<\/th>\n Plugins<\/th>\n Integrated<\/th>\n<\/tr>\n<\/thead>\n \n No<\/td>\n Standard APIs for various services<\/td>\n This comparison highlights the specificities of each framework, as well as how they align with the growing needs of a cloud-native future. Dapr: A New Framework for Orchestrating AI Agents<\/td>\n On March 12, 2025, the Dapr project<\/td>\n<\/tr>\n \n , a microservices orchestration system, presented its new initiative: Dapr Agents. Unlike other frameworks, Dapr focuses on integrating resilience and observability into its workflows. Roberto Rodriguez, one of the core maintainers, explained the motivations behind the creation of this extension.<\/td>\n Key Features of Dapr Agents<\/td>\n Dapr Agents, based on Dapr Workflow, offers a robust architecture for managing AI agents. This system integrates crucial elements to ensure seamless interaction between agents and the microservices infrastructure. Highlights include:<\/td>\n Long-Running Processes:<\/td>\n<\/tr>\n \n Agents are long-running processes that handle multiple tasks simultaneously.<\/td>\n Avoidance of Rigid Coding:<\/td>\n Users can choose and adjust the right tools for each task without having to reprogram the entire workflow.<\/td>\n Multilingual Support:<\/td>\n<\/tr>\n<\/tbody>\n<\/table><\/figure>\n\n Impact on Microservices Architecture<\/h2>\n\n
Dapr Workflow<\/h3>\n\n
Standard<\/h3>\n\n
\n\n
\n \nSynergies between SKPF and Dapr<\/th>\n SKPF aims to enhance scalability while providing integrated orchestration within workflows. This rapprochement with Dapr could allow better use of existing infrastructure, although voices are being raised to demand greater collaboration between the two communities.<\/th>\n It should be noted that the compatibility shown between AutoGen and Dapr could lead to increased synergy between the different open source tools. Indeed, this could minimize redundancies and maximize opportunities for innovation. Here is an overview of this dynamic:<\/th>\n<\/tr>\n<\/thead>\n \n Interoperability:<\/td>\n Facilitates the integration of agents developed under AutoGen and those provided in Dapr.<\/td>\n Simplifying workflows:<\/td>\n<\/tr>\n \n Reduces complexity within systems by merging best practices.<\/td>\n Accelerates time to market:<\/td>\n Contributes to the rapid deployment of AI solutions in diverse environments.<\/td>\n<\/tr>\n \n Future Market Analysis: Outlook and Challenges<\/td>\n In light of the innovations presented, it is undeniable that the landscape of artificial intelligence and AI agents is undergoing a transformation. However, challenges remain. The issues of security, data management, and harmonization between different standards are crucial.<\/td>\n Thus, while projects like Kagent and Dapr are shaping a fascinating future for AI, their success will also depend on how the community manages these challenges. Collaboration will be essential in finding sustainable and effective solutions.<\/td>\n<\/tr>\n<\/tbody>\n<\/table><\/figure>\n\n <\/h2>\n\n
<\/h3>\n\n
<\/h3>\n\n