How to use AI to plan a team offsite that works
Atlassian’s Teamwork Lab developed an AI agent to improve team offsite planning by analyzing 69 gatherings and identifying five activity types that drive measurable connection gains.
Team offsites remain essential for modern organizations, yet planning them is costly and complex. Atlassian’s Teamwork Lab studied 69 engineering gatherings between August and mid-October 2024 to determine what makes them successful. Researchers analyzed event size, duration, agendas, and post-event survey responses to identify patterns linked to higher impact. The findings were used to create IRA, an AI agent built in Rovo that helps planners design agendas grounded in research-backed activity types.
The AI agent, IRA, serves as both a strategic guide and administrative assistant for Atlassian teams. It is accessible today within Rovo or via Slack’s #help-itg channel for Atlassian employees. For external users, Atlassian provides a prompt template to replicate the agent’s planning framework. The agent helps classify activities into five research-backed types to ensure gatherings align with proven outcomes.
Atlassian’s research found that in-person gatherings increase team connection by 27% on average, though the effect diminishes over about four months. This pattern, called “connection decay,” suggests a recommended cadence of roughly three gatherings per year. The most effective gatherings set clear goals, include team-building and planning time, and ensure decisions lead to actionable next steps.
The Teamwork Lab’s methodology involved cross-analyzing agenda activities with post-event survey responses from 51 agendas and 56 surveys. The AI agent reduces administrative burden by automating agenda design while keeping it anchored to research-backed best practices. Atlassian emphasizes that the agent supports planners but does not replace human judgment or the collaborative work teams attend gatherings to accomplish.