Kakao has released the 3B-scale multimodal decision-making model Kanana-2-v-decide as open source.
The migration was verified by confirming zero field mismatches in a full reconciliation conducted 7 days prior.
The Kakao Recommendation Team applied Data Catalog and Analyst Persona to unify AI analysis standards.
The previously unobservable international MO segment was converted to a pre-allocation, post-observation structure, securing a conversion rate in the 90% range.
Kakao has released the VODs of key sessions from AI Sail Summit 26, held in Busan on September 29.
Kanana-2-vision-encoder recorded the global highest scores in 8 out of 9 domains.
All sessions of if(kakao)26, held on October 13–14, will be available for online viewing by anyone.
Prevent unnecessary context accumulation and improve exploration efficiency through progressive disclosure and sub-agent utilization
Kakao introduced a personalized ranking model combining GDCN and MoE, improving the valid play rate by 20%.
Kakao has introduced interest-based networking and deep-dive speaker dialogue programs at if(kakao)26.
Automatically provisions an isolated Airflow environment within 5 minutes upon PR creation, supporting Jupyter and VS Code for Web.
Kakao integrated LM-SPT into Kanana-o, improving voice efficiency and style control, achieving 94.50 on the Korean InstructTTSEval benchmark, close to Gemini-2.5-flash-preview-tts's 95.38.
Kakao and Samsung Electronics held a joint hackathon to solve AI livelihood issues as the first collaboration between their digital talent development programs.
Automated analysis procedures by integrating skills and context documents into the existing Hadoop environment without a new platform.
This article details the process of using Google Sheets as a database and defining repetitive tasks as skills to delegate to AI agents.
Ten winning MCP servers have been officially registered on PlayMCP, and Kakao has announced the second competition, 'Agentic Player 10,' for June.
We automated the entire stress testing process, from build to report generation, using Claude Code and Codex.
Supports real-time conversation and high-quality TTS simultaneously through OpenAI API compatibility and a process isolation architecture.
An algorithm was applied to optimize the width and position of booking blocks by calculating the maximum distance between nodes using DFS.
The Kanana model achieved superior performance with 11T tokens, half the amount of global targets, and Kakao is reviewing industry-academia collaborations including GPU resource support.
This covers AI-based monitoring that reduces noise and identifies real threats from vast security signals.
A developer's journey of growth from DB and security to AI, built on reasonable choices rather than right answers.
Through three onboarding stages, new crews learned the standards of server development through TDD and refactoring.
This is a post explaining the problems from the 2026 Kakao Group New Crew Recruitment Coding Test Round 1.
AI Briefing