The same program, twice
Both examples under the Agent node implement one supervisor and two specialists: the LangGraph supervisor and the Agents SDK version. Reading them side by side is the fastest way to feel the difference.
Capability by capability
| Capability | LangGraph | Agents SDK |
|---|---|---|
| Control flow | Explicit graph, edges in code | Implicit, driven by hand-offs the model chooses |
| Determinism | Routing can be plain functions | Routing is a model decision |
| Tracing | Bring your own, OpenTelemetry fits | Built in, provider dashboard |
| Provider coupling | Neutral | Leans OpenAI |
| Testing | Unit-test nodes and edges | Test tools; routing needs evals |
| Checkpointing | First class | Minimal |
Where each one hurts
LangGraph asks you to think about state shape before you have a product. The Agents SDK asks you to trust model routing before you have evals.
LangGraph for anything with more than two hand-offs or a human approval step. Agents SDK for internal tools where a wrong route costs minutes, not money.
Key takeaways
- Pick LangGraph when routing is the hard part and you need to review control flow in code.
- Pick the Agents SDK when you want the smallest surface and built-in tracing on one provider.
- Neither solves permissions. Both need a gateway in front of tools.
More on these topics
Checklist · · 24 checks
Working with Claude, practices that hold up
Prompting, agents and tools, Claude Code, evaluation and safety. The habits that make Claude-based systems reliable, as a checklist you can run against your own setup.
Architecture pattern · · 1 min read
Pattern: the outbox for agent actions
Agents that write to systems of record need the same transactional outbox that event-driven services use. This entry covers the shape and the trade-offs.
Architecture pattern · · 1 min read
Designing a Multi-Agent System for Real-World Use
How to design and run a multi-agent system in production, using the supervisor pattern, with its trade-offs and where human approval belongs.
Discussion