Research a topic, then audit the sources

A commenter uses one agent to gather initial research, then a second to check each reference and point out problems.

Done with Manusยท Intermediate

What they did
They start by asking Manus to do the first round of topic research. After that, they paste the output into Genspark and tell it to review every citation, check support for each claim, and call out any other issues. The same commenter also has the AI write in one of several identity styles they keep for themselves.
How it went
They say this workflow has gone well so far. They also feel both tools make up less than other models, though that is just their personal impression.
Tips
This setup works best when you separate gathering from checking. Before using anything that could be sent, published, or finalized, have the agent pause and wait for your approval.

Try it yourself with Manus

Research [topic] for a [length] overview focused on [key questions]. For each claim, include the source and the exact point it supports. Then review your own draft for weak spots, missing support, or likely errors. Stop and ask me to approve before you send, submit, publish, delete, or otherwise finalize anything. If a claim cannot be backed up, flag it clearly so I can verify it.

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

Read the original on johnquigginblog.substack.com โ†—

Get 3 agent uses like this every morning

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