AI workflow to replace your morning news scroll

Executive overview

Staying current with AI research, industry news, and key voices is impossible to do manually at scale. Morgan Brown, VP of Product at Dropbox, solved this by building a daily AI workflow that scans papers, newsletters, X accounts, podcasts, and YouTube — then surfaces only the top signals ranked by importance.

Build the prompt conversationally first, refine the output, then schedule it to run automatically.

Building the workflow

• Start by listing the sources you already check manually (arXiv, Substack, X accounts, podcasts)
• Feed those into a prompt via voice or text — stream of consciousness works
• Ask the AI to expand beyond your seed list to adjacent sources
• Define a ranking scale: world-changing, interesting, incremental, noise
• Request synthesis in the context of your specific role, not just a raw list
• Ask the AI to structure the conversation into a clean, reusable prompt
• Review and iterate — treat the prompt like copy, refining until the output is right

Running and scaling it

• Run the prompt once and evaluate the output — adjust length, sources, and balance
• When satisfied, schedule it as a daily task (e.g. 5am) so results are ready with your coffee
• Morgan also built a parallel agent to manage personal Gmail: categorise, summarise newsletters, flag priority emails, draft responses