A client wants to pull fields from invoices with one known layout, and the steps never vary. They ask for a multi-agent system. What do you recommend?
Why A is the answer
A single-step workflow with a schema and validation
Match the pattern to the problem. Known steps need no planning, and each extra agent adds cost, latency and failure modes with nothing gained here.
In detail
- One layout and fixed steps need no planning.
- A schema plus validation gives a reliable, testable result.
- Each extra agent adds cost, latency and failure modes, with nothing gained here.
- Show the client the cost and failure points of each extra agent.
Use case
An illustrative scenario, not a real incident.
A client wants invoices with one layout turned into records. One call with a schema, then validation in code, handles it. The proposed team of agents would add cost and nothing else.
See it in different ways
flowchart LR A["Single call with retrieval"] --> B["Workflow"] B --> C["Single agent"] C --> D["Multi-agent system"]
stateDiagram-v2 [*] --> SingleCall SingleCall --> Workflow: steps are known Workflow --> Agent: path depends on findings Agent --> MultiAgent: evals show overflow or specialist failures MultiAgent --> [*]
flowchart LR S0["Invoice"] --> S1["Claude with a schema"] --> S2["Validation in code"] --> S3["Record stored"]
Key point: Known steps need no planning.
Why the other options are wrong
- B. An autonomous agent that plans each document
- Planning adds nothing when each document follows the same layout.
- C. A manager agent with an evaluator subagent
- A manager and an evaluator add two layers to a single-step job.
- D. A coordinator with one subagent per field
- One subagent per field multiplies calls for a job that one call can do.