Talk title: From Workflows to multi-agent automation: The logical progression of AI in biology research
Assistant Director for Shared Resources, Stephenson Cancer Center, University of Oklahoma Health, Oklahoma City OK 73104.
Artificial intelligence is reshaping how research gets done — but "using AI" is not a single skill. It is a spectrum of autonomy, and knowing where a given task belongs on that spectrum is what separates useful automation from wasted effort and avoidable risk.
This presentation provides a practical map of that spectrum through four stages of increasing autonomy. We begin with AI as an assistant, where the model accelerates individual tasks — summarizing literature, drafting protocols, debugging analysis code — while you stay in control of every step. We then move to workflow automation, using no-code tools such as n8n and Power Automate to let fixed, repeatable pipelines run themselves. From there we cross a crucial threshold into agentic AI, where the model is given a goal rather than a script and decides its own steps. Finally, we reach multi-agent systems, in which specialized agents coordinate as a team to tackle larger research tasks such as automated literature reviews.
