AI Briefing#kakao
The latest AI and developer news about #kakao, with the original source and a short summary.
The latest AI and developer news about #kakao, with the original source and a short summary.
Kakao has open-sourced three models under Apache 2.0 that detect harmful content, legal risks, and prompt attacks.
AI reduced deployment note generation time by 80% and automated risk analysis based on past incident cases.
Ensures stability and security through alternative model chaining during LLM failures and tokenization of sensitive data.
Kakao validated the possibilities and limitations of AI collaboration by building an AI-based hackathon judging system in just 48 hours.
Introduces an Agentic Coding guidebook along with a cheat sheet for AI collaboration and a standard ruleset sharing system.
Kakao's level-based customized AI education resulted in over 60% of all tech roles participating in the AI Basic course and over 60% of all developers acquiring generative AI application capabilities.
Kakao is fostering an experimental culture within the organization that does not fear failure by hosting AI Beer Chat Day, where AI agents are utilized as team members.
Pilot results from Tech Strategy's AI Milestone program show that 68.4% of developers find it difficult to develop without AI tools, while the proportion experiencing time savings through AI use in the design phase rose by 21.2%p over three months.
Kakao's open-source automation management system 'OLIVE' reduced manual mapping items by 95% and cut average license review time from 60 minutes to under 10 minutes through the introduction of AI agents.
Ambassadors will serve for three months starting in October, enjoying early access to new products and offline meetup benefits.
Kakao Makers built a process-level monitoring pipeline using PM2, Grafana Alloy, and TSCoke to overcome the limitations of OS-level monitoring for Next.js SSR servers.
Kakao has released the beta version of 'PlayMCP', a developer platform that supports Remote MCP server registration and AI chat testing.
Confirmed average 50% to maximum over 100% productivity improvement in collaboration environments based on legacy code
The conference will be held at the Kakao AI Campus in Yongin from September 23 to 25, with registration opening on August 28.
AIME benchmark performance was improved through high learning rates in the SFT stage and relaxed clipping strategies in RL.
AI and cloud technology seminars and hackathons from major companies including Kakao, SKT, and AWS are concentrated from late August through September.
By applying the Kanana model, it achieves a 98% receipt recognition rate and automatically recommends settlement categories.
Three backend developers won by completing a Chrome extension in 10 hours using vibe coding.
Kakao has revealed a Spark job design featuring four-stage modularization and design patterns, including Reader and Translator, to ensure data reliability across approximately 300 CDC pipelines.
Kakao has released the lightweight multimodal language model 'Kanana-1.5-v-3b' as open source for commercial use, achieving superior performance compared to competing models on Korean benchmarks.
Kakao has open-sourced 'Kanana-1.5-15.7B-A3B', a model utilizing an MoE structure to achieve performance comparable to Dense 8B models with only 3B active parameters.
Kakao has validated an average 2x productivity improvement and production-grade code quality through the adoption of AI coding agents.
Using vibe coding, an AI judging system for a 250-person hackathon was built in just 48 hours.
Development time was reduced to 10 hours and AI judging was introduced, with 250 participants from all job roles.