> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cohesive.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Research a list in parallel

> Gather comparable evidence for a list of companies, vendors, or other independent items.

Give the agent one shared rubric and a list of independent items to research, then compare the reviewed findings.

<Steps>
  <Step title="Add the list and criteria">
    Put a spreadsheet or note on a Canvas with the items to research. Add the questions you need answered and any sources that should take precedence.
  </Step>

  <Step title="Request independent profiles">
    Select the sources and press `C`.

    *"Research each company independently. Create a profile with what it does, customer segment, public pricing where available, and dated source links. Mark unsupported fields as unknown."*
  </Step>

  <Step title="Inspect progress and evidence">
    For a Canvas task batch, open tasks from the process tray to inspect their work. Check source dates and review a sample before asking for the final comparison.

    Ask for retries only where results are incomplete or failed.
  </Step>

  <Step title="Combine the findings">
    Ask: *"Compare the reviewed profiles against our criteria. Explain the tradeoffs and keep missing evidence visible."*
  </Step>

  <Step title="Keep the reusable context">
    Organize profiles and the comparison into sections. Add durable files to the Workspace and record the research date for the next review.
  </Step>
</Steps>

Parallel work can shorten the wait, but duration depends on research depth, concurrency, and provider limits. See [Parallel tasks](/product/agent/parallel-tasks).

The same method works for vendor proposals or account research. Use a specific rubric and verify the evidence before making a decision.