AI Briefing#ai-agents
The latest AI and developer news about #ai-agents, with the original source and a short summary.
The latest AI and developer news about #ai-agents, with the original source and a short summary.
The Kakao Pay team used AWS AI services to develop the 'Geum.Yong.Sa.' service, which explains difficult financial terms and recommends related products.
Replit is introducing an **Effort-Based** pricing model that charges based on the complexity and time required for a task.
Replit has launched Replit Import, a feature that converts designs from Figma, Lovable, Bolt, and more into apps.
Holo1, an open-source VLM family specialized for GUI automation, and the web agent Surfer-H have been released.
It presents a method of combining CodeAgent with structured JSON output to enhance the reasoning and execution reliability of agents.
Mistral has announced the Agents API for building AI agents equipped with code execution, web search, and image generation capabilities.
Replit has unveiled Element Editor, which lets users intuitively edit text and styles like a design tool.
Hugging Face has released Tiny Agents for Python, which supports the MCP protocol for easy tool use.
OpenAI has unveiled **Codex**, a cloud-based agent that performs software engineering tasks in parallel.
LangChain officially launched the LangGraph Platform at Interrupt 2025 while emphasizing the importance of agent engineering.
Expedia is pursuing marketing innovation through AI, addressing data analysis, content creation, and changing consumer behavior.
Mistral AI has unveiled **Le Chat Enterprise**, an enterprise AI platform based on Mistral Medium 3.
Gradio apps can now be turned into MCP servers callable by LLMs with just a few lines of code.
This introduces a way to implement an ultra-lightweight AI agent in just 50 lines of code using MCP.
This article explains the concept of vibe coding proposed by Andrej Karpathy and the importance of prompt engineering, comparing the features and use scenarios of four AI coding tools: Cursor, GitHub Copilot Agent, Claude Code, and OpenAI Codex CLI.
Highlights how the quality of outputs has surged without changes in prompting skills due to advancements in LLM and agent technologies.
Building reliable agentic systems requires understanding the difference between workflows and agents, and controlling the appropriate context.
OpenAI has unveiled its next-generation models o3 and o4-mini, featuring powerful reasoning capabilities and agentic functionality.
By leveraging Replit Agent to build its own QA tool, Zinus saved over $140,000 in development costs and reduced development time by 50%.
OpenAI has launched the GPT-4.1 model series via API, featuring greatly improved coding, instruction-following, and long-context understanding capabilities.
A new benchmark called **BrowseComp** has been released to measure AI agents' ability to deeply explore information.
Anthropic is hosting its first developer conference, 'Code with Claude,' covering hands-on implementation cases using the Anthropic API and MCP.
GitHub CEO Thomas Dohmke visited KakaoTech to share Copilot's 2025 roadmap and AI agent strategy, discussing developer competencies in the AI era.
OpenAI has released PaperBench, a benchmark that evaluates AI agents' ability to reproduce state-of-the-art AI research.