When do several agents help, and when do they just add problems?
Answer out loud or in your head first, then compare.
Answer
Several workers help when the work can be split into independent parts. Each gets a clear goal, the data it needs and the expected result. For example, three different sources can be analysed separately and the conclusions compared afterwards.
If they all keep changing the same file or need every piece of information from the others, you add coordination costs and conflicts. More agents also means more model calls. Several models voting doesn't prove the truth, especially when they use the same faulty data. I compare time, cost and quality against a single worker. In a short interview task, one well-run session may be more sensible.
Why it matters: Lets you justify parallelism without treating it as a goal in itself.
Possible follow-ups: Who merges the results, and what do they do with a disagreement?
How did it go? Be honest — “Not yet” puts the question on your review list.
Start with
- H-P01 · How does an AI agent differ from a plain model call? — Agents and workflows
- H-P02 · What is a workflow, i.e. a fixed sequence of work? — Agents and workflows
Further reading
- Anthropic: How we built our multi-agent research system(opens in a new tab)13.06.2025. Coordination, costs, and tasks that suit parallel work.