🏆 OfficeOS Wins First Place at the Agent Glow Up Hackathon

<figure><img height="853" width="1280" src="https://substackcdn.com/image/fetch/$s_!Vwih!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fc01976bf-3801-4d0f-815d-28b8774cb90b_1280x853.jpeg"></figure><p>Instead of hiding AI behind chat windows, OfficeOS puts each business department front and centre, with agents that are visible, manageable, and ready to work.</p><p>Check the demo below:</p><p>👉 <a href="https://campus.buildclub.ai/projects/019ccae3-63ac-7dc1-963a-e6b4a53982ac">Explore OfficeOS here</a></p><hr><figure><img height="679" width="1456" src="https://substackcdn.com/image/fetch/$s_!5UWs!,w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F1801f733-7fc3-4ddd-8f0a-ff2bce5f3b7b_3364x1568.png"></figure><h3>The Project: OfficeOS</h3><p>OfficeOS reimagines what it means to delegate work - to AI.</p><blockquote><p>“Instead of interacting with AI through chat windows or isolated tools, I built an office where each department - sales, operations, legal, and finance - is an agent you can assign work to.”</p></blockquote><hr><h3>Why the ‘Agent Glow Up’ Hackathon?</h3><p>The hackathon’s theme wasn’t just a fit - it became the forcing function for the core idea.</p><blockquote><p>“The hackathon pushed me to focus less on individual agent intelligence and more on the experience of managing agents - which ultimately became the core idea behind the system.”</p></blockquote><hr><h3>Challenges</h3><p>Scoping a multi-agent system in four hours meant making hard calls about what to actually build.</p><blockquote><p>“The biggest challenge was deciding what to implement versus what to simulate - the concept involves so many interacting components.”</p></blockquote><p>Integrating Prefect to coordinate the research agents - while keeping the demo visually clear - required learning new tooling quickly and iterating under real pressure.</p><hr><h3>Highlights</h3><p>Framing AI agents as office departments - not chat windows - made the whole system instantly click for people.</p><blockquote><p>“I discovered how effective a visual metaphor can be - representing AI systems as departments in an office made complex workflows easy to understand and explain.”</p></blockquote><hr><h3>What’s Next?</h3><p>The goal is to connect OfficeOS to live data so agents can act on real information.</p><blockquote><p>“The next step is deeper integration with real data sources like email, calendars, and financial tools so departments can operate on live information.”</p></blockquote><hr><h3>Advice for Future Hackathon Builders</h3><p>One clear idea beats ten half-built ones.</p><blockquote><p>“Focus on communicating one clear idea rather than trying to build everything - a strong concept that’s easy to demo often resonates more than a technically complex system.”</p></blockquote><hr><h3>Big Picture: Managing Agents Is the Next Frontier</h3><blockquote><p>“The biggest lesson from the hackathon was that the next frontier for AI systems may not just be smarter agents, but better ways for humans to manage and collaborate with them.”</p></blockquote><p>The race isn’t just to build more powerful agents. It’s to build better ways for people to work with them.</p><hr><p><strong>📌 Connect with the <a href="https://www.linkedin.com/in/michael-shears">Michael</a> on LinkedIn</strong><br><strong>📌 Explore their project, <a href="https://campus.buildclub.ai/projects/019ccae3-63ac-7dc1-963a-e6b4a53982ac">OfficeOS</a>.</strong></p><hr>