A 4-Agent Data Analyst Team Lost to One Agent With Retries: 3.5x the Cost on a 50-Question Test
- Level Advanced
- Setup1–3 hrs (estimated)
- Evidence Some evidence
- CostNot stated in source
- No-codeNo
What they did
A developer built a 4-agent data-analysis team (planner, coder, reviewer, explainer) and tested it on 50 questions over three public datasets against two single-agent setups. One agent that got its Python errors back and retried up to 3 times scored 100%. The team scored 48 of 50 after a scorer fix, used 3.6x the tokens, cost about 3.5x as much and was slower (about 4.0s vs 2.4s). One miss came from the explainer agent losing a correct result. All three setups refused all 4 unanswerable questions.
How it went
dev.to post by Dhruvi Rohilla (Oct 29, 2026) with an open-source GitHub repo: 50 questions, single agent with retries 100%, 4-agent team 48/50 at ~3.5x cost. One person's benchmark on simple datasets with one model. Run your own small test before picking a design
Worth knowing
- dev.to post by Dhruvi Rohilla (Oct 29, 2026) with an open-source GitHub repo: 50 questions, single agent with retries 100%, 4-agent team 48/50 at ~3.5x cost.
- One person's benchmark on simple datasets with one model. Run your own small test before picking a design.
Prompt
Cleaned version rewritten from the source — not a direct quote.
Answer this question about my data: [question]. You'll get the column names, types and a few sample rows of [file]. Write Python to answer it, run it, and if it errors, read the error and fix it, up to 3 tries. Give me the exact number plus one plain-English sentence. If the data can't answer the question, say CANNOT_ANSWER and why instead of guessing.
Recommended
Try it in Grok Bot
For Grok Bot users the lesson is the same: one bot that checks its own work often beats a team of bots passing notes.
- Copy the prompt (button below).
- Open Grok Bot chat and paste it. Replace any [bracketed] placeholders.
- Connect the listed tools (as needed) as connectors when Grok Bot asks.
- If the task repeats, ask: "Make this a daily routine."
Alternatives: the same prompt works in Claude, ChatGPT Agent, OpenClaw.
Set it up
Stack: Custom agent
- Read the original source for full context.
- Connect the tools your agent needs.
- Paste the prompt and replace [bracketed] placeholders.
- Run once manually and check the output before scheduling it.
Generic suggested steps, not from the source.
Evidence
dev.to post by Dhruvi Rohilla (Oct 29, 2026) with an open-source GitHub repo: 50 questions, single agent with retries 100%, 4-agent team 48/50 at ~3.5x cost. One person's benchmark on simple datasets with one model. Run your own small test before picking a design
Source: Web · 2026-10-29 · Verified by us: not yet · Report broken or outdated