# Multi-Agent Systems İlgili Makaleler

HTX Haber Merkezi, kripto endüstrisindeki piyasa trendleri, proje güncellemeleri, teknoloji gelişmeleri ve düzenleyici politikaları kapsayan "Multi-Agent Systems" hakkında en son makaleleri ve derinlemesine analizleri sunmaktadır.

From TPU to Self-Evolving Agents: How Jeff Dean Predicts the Next Step in AI

At the 2026 YC Startup School, Jeff Dean outlined his vision for AI's next phase, shifting focus from simply scaling models to building intelligent, autonomous systems. He believes AI's progress is no longer just about creating smarter models, but about integrating them into systems capable of long-term, iterative work, automated experimentation, and continuous learning. This evolution moves the competition from "who has the bigger model" to "who can best organize intelligence." Dean suggests AI capabilities are now comparable to a junior engineer, enabling the automation of complex workflows. However, the true challenge and opportunity lie in managing these AI "workers" at scale. He emphasizes the importance of **context engineering**—structuring tools, memory, and feedback loops—over raw model power. For startups, this means building deep expertise in niche domains where general models currently fail (near 0-1% success rates), leveraging proprietary data, specialized tools, and domain-specific evaluators. A recurring theme is re-examining fundamental constraints. Dean's past work, like moving Google's search index to memory or creating the TPU, stemmed from questioning outdated assumptions about hardware and cost. He sees similar inflection points today, particularly in **specialized inference hardware** to drastically reduce latency and energy consumption for real-time Agent operation. Notably, he points out that in modern AI systems, the dominant cost is often not computation but **data movement**. Reliable, long-running Agents require robust system design, borrowing concepts from distributed computing like checkpointing, state management, and parallel exploration to handle failures and maintain progress over days or weeks. As AI automates execution, the scarcest human skills will shift to **defining clear specifications**, **judging what problems are worth solving** (taste), and designing effective feedback loops. Ultimately, Dean's framework prioritizes understanding the problem deeply, identifying the true bottlenecks, and systematically building closed-loop systems where AI can not only perform tasks but also improve AI itself.

marsbitDün 00:13

From TPU to Self-Evolving Agents: How Jeff Dean Predicts the Next Step in AI

marsbitDün 00:13

Capturing 15 Top-Tier Zero-Day Vulnerabilities: A Consensus Protocol Debug Agent Framework Built by 0G Lab in Collaboration with Teams from NUS, PKU, and BUPT

"Agents Capture 15 Critical Zero-Day Bugs: 0G Lab's Multi-Agent Framework Automates Debugging in Consensus Protocols" Distributed consensus protocols are notoriously difficult to debug due to complex, intertwined states. A novel framework, Agora, developed by 0G Labs with researchers from NUS, Peking University, and Beijing University of Posts and Telecommunications, tackles this by fusing deep domain expertise with a collaborative multi-agent LLM architecture. Agora moves beyond the limitations of single LLMs and traditional testing like fuzzing. It employs three specialized agents: an Orchestrator for global state, a Strategy agent for generating attack scenarios using distributed systems knowledge, and a TestGen agent that creates executable tests. A core innovation is its efficient "Succinct Memory & Communication" mechanism and a dynamic test harness. This allows the system to translate abstract hypotheses into concrete tests across languages like Go and Rust, run them, capture failures, and refine the approach in a closed loop—all with minimal token overhead. In rigorous evaluations on production-level protocols including Raft, EPaxos, and components from etcd and Sui, Agora discovered 15 previously unknown deep logic bugs (e.g., execution divergence, liveness violations). In stark contrast, powerful standalone LLMs like GPT-5.2 and Claude 4.5 found zero such bugs. Agora achieved this with a high precision of 73.9% and at an average cost of only about $40 per bug found. The framework demonstrates high generalizability. Its decoupled design allows the "Multi-Agent + Hypothesis-Driven Testing" paradigm to be applied to other complex domains like database concurrency control, OS kernels, and Web3 smart contract auditing. By enabling efficient, automated detection of deep logic flaws, Agora points the way for AI-powered security in critical infrastructure, aligning with the growing trends of agentic systems and automated quality control.

marsbit06/11 07:12

Capturing 15 Top-Tier Zero-Day Vulnerabilities: A Consensus Protocol Debug Agent Framework Built by 0G Lab in Collaboration with Teams from NUS, PKU, and BUPT

marsbit06/11 07:12

活动图片

Bitcoinright-arrow

Technology Trendsright-arrow