Framework 01
LangGraph
Stateful, graph-based orchestration. Complex workflows with branching logic, loops, and human-in-the-loop interrupts.
StatefulConditionalHITLPersistence
Framework 02
CrewAI
Role-based collaborative agent teams. Each agent has a role, goal, and backstory. Best for specialised multi-expert tasks.
RolesCollaborationDelegationSpecialisation
A1PHASE 2
Multi-Agent Translation Reviewer
Five-agent CrewAI crew. Measurably better quality than single-LLM translation. Each agent specialised for one task.
Translator → Grammar → Fluency → Scorer → Formatter
CrewAILangGraphRagas
A2PHASE 2
Telecom Incident Analysis Agent
Upload any telecom log or CDR. LangGraph detects patterns, finds root cause, generates structured incident report.
Parser → Detector → RCA Agent → Remediation
LangGraphCDRLangSmith
A3PHASE 3
Research Multi-Agent Assistant
Automated research pipeline. Search, extract, evaluate sources, synthesise, and cite — faster than manual research.
Search → Extract → Synthesise → Cite
LlamaIndexTavilyCrewAI
A4PHASE 3
Autonomous Code Review Agent
Submit a PR. CrewAI crew reviews for security, bugs, style, test gaps, docs — structured review with severity scores.
Security → Bugs → Style → Tests → Docs
CrewAIGitHub APILangSmith