AI in Production: Why Multi-Agent Systems Are Moving to Plain Python
A LangChain-in-production teardown: the author built a multi-agent system in plain Python and shows where the framework starts to get in the way. The main failure points are model switching, fallback to YandexGPT, RAG and tool calling. The conclusion is blunt but practical: for a live service, transparent contracts, validation and control over every step matter more than a ready-made universal wrapper.
AI-processed from Habr AI; edited by Hamidun News
Deploying AI in production systems takes a specific approach: Habr AI examined why multi-agent systems are moving away from LangChain in favor of plain Python.
Why doesn't LangChain always work in production?
According to the article, LangChain's universal LLM abstractions do not save time and they undermine predictability in real services. The promise of "switching models with one line" almost never works that simply. Even models from the same provider behave differently: the expensive version reliably returns JSON, while the cheaper one changes keys and forgets system instructions.
How do you use AI in production systems without LangChain?
The author describes a plain-Python stack in which the developer has direct control over model behavior instead of relying on framework abstractions.
- Nov 1, 2026Local models lose to Opus and Codex: 0.21 on a MacBook versus 1.00 for Codex
- Nov 1, 2026Adobe's EditAgent judge picks a usable edit 81.3% of the time, versus 74.8% for EditScore
- Nov 1, 202616 models tested on 7 search tools: Exa averaged 10.5 out of 14, SearXNG about 3
- Nov 1, 2026Google's StreamVibe video agent beats other models by 10.7% on response timing and quality
Key Questions Answered
Why doesn't LangChain always work in production?
According to the article, LangChain's universal LLM abstractions do not save time and they undermine predictability in real services. The promise of "switching models with one line" almost never works that simply. Even models from the same provider behave differently: the expensive version reliably returns JSON, while the cheaper one changes keys and forgets system instructions.
How do you use AI in production systems without LangChain?
The author describes a plain-Python stack in which the developer has direct control over model behavior instead of relying on framework abstractions.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.