Last week at Team'25 Europa in Barcelona, the Rovo breakout sessions by Jensen, Sherif, and also the awesome live demo by Ali Bencafu completely inspired me to rethink my Rovo custom agent design.
What I built into "Atlassian Sherpa" (my custom agent for Atlassian support and extended search, including internal knowledge and governance):
🧠 Behaviour: new Rovo Studio feature. I've moved the role and tone out of the instructions into the Behaviour section. Defined the Sherpa's personality, communication style, and interaction techniques that make it actually feel like a helpful teammate that guides users, not just a cold search interface.
📊 Visual Complexity Framework Emoji-based difficulty levels (🟢🟩🟨🟧🔴) help users instantly understand if they can self-serve or need escalation.
👥 Role-Based Ownership Mapping Clear guidance: "You can handle this on your own" vs "This needs your Org Admin" - protecting specialist capacity while empowering users.
📚 Dual-Resource Intelligence Every response includes both Atlassian Support documentation AND our internal resources (including governance, available portfolio and customisation) - because real solutions live in both places.
Expected impact:
The Team'25 Europe sessions at Barcelona made me expand my curiosity about Rovo agents: they should not be just about answering questions - they're about changing how your organisation relates to your platforms and how users learn async and on their own, no matter where/when.
Israel
0 comments