From Prompt to Multi Agent System: The evolution of our AI product
summary
Qodo describes moving from a single IDE coding agent to an agentic system and now a multi-agent code review that runs in a Python backend and serves enterprise customers (Nvidia among them). Implementation details: built-in MCP transport to avoid requiring Node on JetBrains machines, LangGraph for agent orchestration and observability, and a diamond-shaped multi-agent review pipeline (context collector, parallel specialist agents, shared tools, then an LLM judge) that reduces noise and improves review accuracy in production.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 3:39AI code review product overview
- 5:23First AI agents identified
- 10:06MCP for dynamic context
- 14:27LangGraph workflows
- 16:56From opinionated to React agents
- 22:22Multi-agent system expansion
- 24:29Architectures for multi-agent review
- 27:01Diamond-shaped architecture design
- 31:18Staying relevant in AI
- 35:10Q&A
related talks
25mFrom AI Chaos to Production Ready: Using AI Agents to Improve Your Code Quality
40mHow To Tame Your Agents?
25mBeyond the IDE: Toward Multi-Agent Orchestration
29mPairing with the machine: How AI Agents are rewriting the rules of software development
35mThe Agentic AI Engineer
16mWhat Comes After AI Coding Assistants
01h 03mAI Coding Agents and how to code them
30mLessons From Building and Deploying a Multi-Agent System at Scale: An...