Use parallel agents to run a research workflow

Sakana’s Agentic Scientist coordinates several agents at once to carry a project from idea to write-up.

Done with Custom agent· Advanced· Oct 2026

What they did
Sakana announced a third release of its AI research tool, Agentic Scientist, and pointed to the project’s GitHub repository. The update describes a shift away from one agent doing the entire paper-writing job and toward a framework of concurrent agents that manages the full research cycle. It does not spell out the individual jobs of those agents or explain where a person would intervene.
How it went
The announcement did not share benchmark results, output examples, or any assessment of the system. It also gave no mechanics, pricing, or limits, saying readers should consult the repository for those details.
Tips
If you want to try a similar setup, ask the agent to map the workflow first and wait for you to approve the plan before it starts. Keep a human checkpoint for anything that could submit, send, delete, or spend money.

Try it yourself with Manus

I’m working on [research question]. Build a workflow with multiple agents running in parallel: one to gather prior work, one to suggest hypotheses, one to test those ideas on [data or sources I provide], and one to draft the final report. Show me the task split and the review points before you begin, and stop for my approval before any step that would submit, send, delete, or pay for anything.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on bsky.app ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.