How one leader rebuilt his feedback loop with an AI agent
Atlassian’s Fernando Garcia Valenzuela built AI agents to audit his Slack messages, identifying missed recognition and tone shifts across a 70+ person, multi-timezone team to refine his leadership communication.
Fernando Garcia Valenzuela, Head of Cloud Storage Engineering at Atlassian, developed a personal AI system to audit his Slack communications after recognizing that empathy and warmth were eroding amid his team’s growth and distributed structure. The experiment used AI agents named Smedley and Brundle to analyze his messages, flag missed recognition opportunities, and suggest rewrites, aiming to maintain his communication standards without sacrificing speed.
Smedley, operating daily, scans direct messages with direct reports to identify missed moments for recognition and proposes warmer alternatives, while Brundle runs biweekly to analyze broader communication patterns across channels and conversations. Both agents were built using existing tools—Slack for data, Rovo for AI processing, and Confluence for reports—demonstrating a portable approach to other platforms.
The agents rely on a structured prompt taxonomy grounded in communication research from experts like Teresa Amabile and Amy Edmondson, providing pragmatic feedback rather than generic advice. Fernando’s system revealed consistent patterns in his messaging, such as overlooked acknowledgments and tone shifts, which shaped whether team members felt seen or overlooked across hundreds of interactions.
Fernando’s experiment evolved beyond audits, with the AI’s presence subtly influencing his real-time message composition and prompting him to add an on-demand editorial module for immediate feedback. Examples from the system show how minor tweaks—like adding five words of acknowledgment—transformed messages from transactional to connective, highlighting AI’s role in refining leadership communication.