https://venturebeat.com/orchestration/new-mit-and-sakana-ai-framework-uses-an-llm-judge-to-cut-evaluation-costs-for-self-improving-coding-agentsVentureBeaten2026-10-02T22:50:08.168+00:00New MIT and Sakana AI framework uses an LLM judge to cut evaluation costs for self-improving coding agentshttps://images.ctfassets.net/jdtwqhzvc2n1/1J7M5sir7llu69q7cACSLv/3c39a40723a1e4a6665cb9ded1dcb812/Branching_self-improving_AI_agent.png?w=300&q=30 https://venturebeat.com/technology/amazon-unveils-a-free-fast-open-source-jev-killer-strands-decider-2b-makes-decisions-in-fractions-of-a-secondVentureBeaten2026-10-01T16:43:09.518+00:00Amazon unveils a free, fast, open source Jev killer: Strands Decider 2B makes decisions in fractions of a secondhttps://images.ctfassets.net/jdtwqhzvc2n1/6nPhWPeaiOYPjrPbqP15KT/50e3cb9087a5de6d49d4eaf3f4c6699c/Broken_Green_Rope_Releasing_Digital_Sparks.png?w=300&q=30 https://venturebeat.com/orchestration/googles-wikiskill-gives-ai-agents-a-memory-of-what-went-wrong-without-putting-it-in-the-promptVentureBeaten2026-10-01T16:42:13.644+00:00Google’s WikiSkill gives AI agents a memory of what went wrong — without putting it in the prompthttps://images.ctfassets.net/jdtwqhzvc2n1/qkeGTjF09KPPhFimb6rgU/b2a4fcd62eeb4215a62a934955a12c59/agent_skill_evolution.png?w=300&q=30 https://venturebeat.com/security/ai-agents-need-more-than-access-control-they-need-identity-at-runtimeVentureBeaten2026-10-01T14:30:00.000+00:00AI agents need more than access control — they need identity at runtimehttps://images.ctfassets.net/jdtwqhzvc2n1/uK7U1cjTR3qQWrtyaZdVj/e2b75550209ad1ce0de936db95a50ce9/Cisco_Duo.png?w=300&q=30