I wanted to start a broader creative experiments thread around Research Agents.
A useful angle is setup choices, but it also connects with common mistakes. Different tools and workflows can produce very different results even when the goal is similar.
It would be interesting to compare...