{"slug": "ai-conf-2026-classic-ml-is-dead-everyone-s-building-agents", "title": "AI Conf 2026: Classic ML Is Dead, Everyone's Building Agents", "summary": "At the AI Conf 2026 in Moscow, the industry has fully shifted away from traditional machine learning toward agents, RAG, and voice systems, with agent orchestration emerging as the new infrastructure layer. A developer built a working ReAct search agent for $0 in prototyping costs, noting that companies now win through superior agent architecture rather than the best single model. The conference highlighted that while voice-to-voice models exist, production stacks still rely on Speech-to-Text → LLM → Text-to-Speech pipelines, and future papers will likely include AI-verifiable code archives.", "body_md": "Spent two days at AI Conf in Moscow. The shift is complete: nobody talks about traditional ML anymore. It's all agents, RAG, and voice systems.\n\nAverage time from submission to publication at A-tier conferences: **9 months.** Multiple review cycles, sequential improvements.\n\n**What researchers actually use LLMs for now:**\n\n**Prediction:** Future papers will include zip archives of experimental code that AI can verify. Human value shifts to idea generation, not implementation.\n\nBuilt a working ReAct search agent in the workshop:\n\nStack cost: $0 for prototyping. Production cost: depends on scale.\n\n| Tool | Approach | Best For |\n|---|---|---|\nLangfuse |\nManual integration, detailed traces | Custom setups, granular control |\nArize Phoenix |\nAuto-instrumentation, wraps everything | Quick setup, less configuration |\n\nBoth show traces, token counts, latency breakdowns. Phoenix wins if you want observability without wiring it yourself.\n\nThe terminology evolved:\n\nExample: Deep Agents framework. Skill creation costs 2M tokens. Single invocation: 100K tokens. But the abstraction is cleaner than manual tool orchestration.\n\nVoice-to-voice models exist but lack:\n\n**Current production stack:** Speech-to-Text → LLM → Text-to-Speech\n\nVoice-to-voice will replace this eventually, but not before tool calling and context compression catch up.\n\nThe industry moved on. If you're still pitching Random Forest improvements, you're talking to the wrong audience.\n\nThe conference confirmed what I see in production: **agent orchestration is the new infrastructure layer.** Not the models themselves — how you connect them, manage memory, route between skills, and monitor everything.\n\nThe companies winning aren't those with the best single model. They're those with the best agent architecture.\n\n*More production AI insights and conference notes from a bank's DS lead — follow my Telegram channel:*\n\n** https://t.me/ai_tablet** (Russian, technical)\n\n*More AI engineering notes, RAG benchmarks, and production insights from inside a bank — follow my Telegram channel:*\n\n🚀 ** https://t.me/ai_tablet** (Russian, technical)", "url": "https://wpnews.pro/news/ai-conf-2026-classic-ml-is-dead-everyone-s-building-agents", "canonical_source": "https://dev.to/__2ddbae6bb7d/ai-conf-2026-classic-ml-is-dead-everyones-building-agents-og6", "published_at": "2026-05-29 06:00:17+00:00", "updated_at": "2026-05-29 06:13:40.416803+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-agents", "ai-tools", "ai-research"], "entities": ["AI Conf", "Moscow", "Langfuse", "Arize Phoenix", "Deep Agents"], "alternates": {"html": "https://wpnews.pro/news/ai-conf-2026-classic-ml-is-dead-everyone-s-building-agents", "markdown": "https://wpnews.pro/news/ai-conf-2026-classic-ml-is-dead-everyone-s-building-agents.md", "text": "https://wpnews.pro/news/ai-conf-2026-classic-ml-is-dead-everyone-s-building-agents.txt", "jsonld": "https://wpnews.pro/news/ai-conf-2026-classic-ml-is-dead-everyone-s-building-agents.jsonld"}}