Practical notes: LangGraph vs. Google ADK in 2026. Part 1: Two Different Ways

The following notes reconstruct a practical path around “LangGraph vs. Google ADK in 2026. Part 1: Two Different Ways to Think About Agent Orchestration”. Emphasis stays on contracts, checks, and drop-in code placeholders rather than motivational framing. When working through the Overview stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.

First: LangGraph and ADK Are Getting Closer

The First LangGraph and ADK stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline. Keep graph state flat and typed. Nested blobs hide which node wrote which field and break resume after interrupts.

The Way you Think About LangGraph

The The Way you Think stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Keep graph state flat and typed. Nested blobs hide which node wrote which field and break resume after interrupts.

State
  ↓
Node
  ↓
Decision
 ↙   ↘
A     B
↓     ↓
Tool  Agent
 \     /
  ↓   ↓
 Validation
     ↓
    End

The Way you Think About Google ADK

The The Way you Think stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Keep graph state flat and typed. Nested blobs hide which node wrote which field and break resume after interrupts. The The Way you Think stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.

Coordinator
├── Specialist Agent
├── Specialist Agent
├── Specialist Agent
└── Specialist Agent

The Most Important Architectural Question: Who Controls the Workflow?

For the The Most Important Architectural stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline. Put human approval on edges that spend money or change production data. Compile-time wiring does not equal business completeness.

Mostly Deterministic Workflow

For the Mostly Deterministic Workflow stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Put human approval on edges that spend money or change production data. Compile-time wiring does not equal business completeness.

Receive request
→ validate
→ retrieve information
→ classify
→ call service
→ validate result
→ human approval if necessary
→ respond

Mostly Agent-Driven Workflow

For the Mostly Agent-Driven Workflow stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Put human approval on edges that spend money or change production data. Compile-time wiring does not equal business completeness. For the Mostly Agent-Driven Workflow stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.

User goal
   ↓
Coordinator
   ↓
Which specialist is needed?
   ↓
Delegate
   ↓
Maybe call another specialist
   ↓
Synthesize

LangGraph’s Biggest Strength: Explicit State

When working through the LangGraph s Biggest Strength stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline. Checkpoint after expensive steps. Resume should not re-bill the same LLM call when an operator retries a later node.

What has already happened?What did the previous agent produce?Which tools were executed?What has been validated?What still needs approval?Where should execution resume?What should the next node actually see?

ADK’s Biggest Strength: Agent Composition

When working through the ADK s Biggest Strength stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Treat this stage as a contract between inputs and validated outputs. Name the artifacts, define success checks, and refuse silent partial completion. Checkpoint after expensive steps. Resume should not re-bill the same LLM call when an operator retries a later node.

                Coordinator
                      │
        ┌─────────────┼─────────────┐
        ↓             ↓             ↓
    Research       Domain        Action
     Agent          Agent         Agent
        │             │             │
        └─────────────┼─────────────┘
                      ↓
                  Validation

Multi-Agent Does Not Automatically Mean Better

When working through the Multi-Agent Does Not Automatically stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Record timings and token or query cost next to functional results. Cost visibility early prevents surprise bills when the path moves from demo to shared environments. Checkpoint after expensive steps. Resume should not re-bill the same LLM call when an operator retries a later node. When working through the Multi-Agent Does Not Automatically stage, write down the contract first: required inputs, success signal, and what happens on partial failure. That checklist keeps later code changes honest. Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.

1 agent = simple
5 agents = sophisticated
20 agents = extremely sophisticated

Part 1 Takeaway

The Part 1 Takeaway stage works best when treated as a measurable surface. Capture one golden transcript, one failure case, and the rollback note before expanding scope. Prefer small, testable units over sprawling scripts. When a step fails, the failure should point at a single responsibility rather than a tangled pipeline. Keep graph state flat and typed. Nested blobs hide which node wrote which field and break resume after interrupts.

Operational checklist

For the Operational checklist stage, define the inputs, the owner of the step, and the exit criteria before changing code. Operators should be able to re-run the step from a known checkpoint without guessing hidden state.

Keep configuration outside application code. Environment files, secret stores, and feature flags belong in one place operators can audit without reading the whole graph.

Put human approval on edges that spend money or change production data. Compile-time wiring does not equal business completeness.

Write a short runbook: how to rotate keys, how to drain the queue, how to roll back the last ingest.

Document the happy path and the recovery path together. Retries, human gates, and dead-letter handling are part of the product, not later polish.

Put human approval on edges that spend money or change production data. Compile-time wiring does not equal business completeness.

Before promoting the stack, freeze versions, capture a golden transcript for the critical path, and confirm rollback steps. Shared environments need rate limits, tenancy checks, and a clear owner for secret rotation. Prefer boring reliability over clever one-off demos.

Batch note for e1d3c18ee0aa: keep provider keys out of the repo, set a per-session token ceiling, and store transcripts next to the eval fixtures so later model swaps stay comparable.