AI Meetup #3 - AI, Agents & demos

For people who build software. We focus on AI, design, engineering, cloud, and startups without the hype.
No vendor pitches or buzzwords. Just real architectures, production lessons, and honest trade-offs. For engineers, architects, platform teams, and technical founders.

For any questions, enquiries, collaborations or partnerships please email : **buildersfoundry@gmail.com**

or connect with hosts - Shubhangi Goyal or Dieter Gobeyn

### Event Agenda

6:00 PM — Doors open
6:30 PM 🤖 1. Supercharge your agents' UX in Microsoft Copilot Studio with Adaptive Cards
Speaker: Angeliki Patsiavou (Microsoft MVP)
7:15 PM — Break (10 mins)
7:25 PM 🤖 2. Your Agent Trace Is Missing the Code: Runtime Observability for Long-Horizon Agents
Speaker: Raluca Crisan (CTO, Co-founder, Etiq AI)
8:30 PM — Wrap up
🍕 Pizza and drinks, as always!

Talk details below:

🤖 1. Supercharge your agents' UX in Microsoft Copilot Studio with Adaptive Cards - Angeliki Patsiavou (Microsoft MVP)
Agent building without considering end-users’ experience is like trying to play Bohemian Rhapsody on a broken guitar—chaotic, messy, and frustrating for everyone involved. Developers often jump in full of enthusiasm, only to face messy outputs, poorly summarized data, and automation flows that leave users scratching their heads. Without a solid design, human-in-the-loop strategies, and testing, your outputs can miss all the right notes.

The key to success? Leveraging Adaptive Cards with clarity, formatting outputs effectively, and keeping humans in the loop for critical interactions. Testing your cards in advance and embedding them seamlessly in Power Apps, Power Automate, or Power Pages ensures your automation flows feel smooth, intuitive, and ready for prime time. With the right approach, your Copilot Studio cards can hit the perfect UX chord every time.

In this session, I’ll guide you through the key elements of Adaptive Card mastery in Copilot Studio. You’ll see live demos, learn best practices for card design and output summarization, explore human-in-the-loop automation patterns, test like a pro, and embed your cards into real applications. Walk away ready to turn your Adaptive Cards into a rock solid, crowd-pleasing experience.

🤖 2. Your Agent Trace Is Missing the Code: Runtime Observability for Long-Horizon Agents - Raluca Crisan (CTO, Co-founder, Etiq AI)
This session presents an execution-level observability architecture for long-horizon, code-driven AI agents, addressing gaps left by conventional agent traces and generic memory systems. The core innovation is the "etiq graph," which is automatically built at runtime to capture the actual relationships between executed code, consumed states, and produced artifacts—complementing the semantic trace of what an agent intended to do with a record of what its code actually did. Using a hands-on multi-stage agent workflow, the session demonstrates a practical development loop: reviewing evidence, retracing a faulty output back to its earliest divergence point, repairing that branch, and resuming only the dependent downstream work rather than restarting entirely. The graph also enables smaller, operation-specific context packages instead of feeding entire histories into each model call. This connects better execution evidence to improved agent performance—reducing duplicated work, preserving valid progress, and shortening debugging cycles. Attendees leave with a reusable pattern for treating the execution graph as both agent memory and a shared debugging pane.