Molecule AI
ArchitectureTechnical Reference

Technical Overview

Executive summary and product positioning for the Molecule AI platform.

Part of the Comprehensive Technical Documentation. Definitive reference based on a non-invasive scan of the molecule-core repository.

1. Executive Summary

Molecule AI is an org-native control plane for heterogeneous AI agent teams. It is not a workflow builder or a replacement for agent frameworks; rather, it is the operational and organizational layer that sits above multiple runtime frameworks and provides:

  • Workspace-as-role abstraction (not task nodes)
  • Hierarchy-driven topology (org chart = communication paths)
  • Hierarchical Memory Architecture (HMA) with LOCAL/TEAM/GLOBAL scopes
  • A2A (Agent-to-Agent) direct inter-workspace communication via JSON-RPC 2.0
  • Canvas-based visual team building with drag-to-nest hierarchy
  • Comprehensive control plane operations — registry, heartbeats, lifecycle, approvals, secrets, traces, bundles

Seven runtime adapters ship production-ready on main: Claude Code, LangGraph, AutoGen, OpenClaw, Hermes, Codex, Google ADK, plus external (bring-your-own agent). The canonical set is the control-plane provider registry.


2. Product Positioning

Core Narrative

One-liner: "Molecule AI is the org-native control plane for heterogeneous AI agent teams."

Five Key Differentiators

#PrincipleImplication
1Workspace = role, not taskInternal AI model can swap, but organizational identity persists across model/framework changes
2Org chart = topologyHierarchy determines communication boundaries — no manual edge wiring needed
3Heterogeneous runtime support7 runtimes shipped; teams choose freely without forced standardization
4Memory follows org boundariesHMA prevents over-sharing, aligns data isolation with organizational structure
5Skill evolution loopmemory → signal → skill → hot-reload → operational improvement (self-improving flywheel)

What Molecule AI Is NOT

CategoryExamplesHow Molecule AI Differs
Workflow buildersn8n, Windmill, TemporalMolecule AI models roles, not tasks
Agent frameworksLangGraph, CrewAI, AutoGenMolecule AI sits above frameworks as runtime adapters
Coding agentsClaude Code, Cursor, CodexMolecule AI runs coding agents as workspace roles alongside other types
Chat UIsChatGPT, Claude.aiMolecule AI is operational infrastructure, not a conversation interface

On this page