Context Engineering Workshop
Prompts tell an agent what to do. Harness and context decide whether it can do it well.
Prompt engineering got your team to its first working agent. But as agents take on real tasks, the same pattern keeps showing up: answers that sound right and aren’t, latency that grows with every tool call, and token bills that grow faster than usage. These aren’t prompt problems: they’re harness and context problems.
In this session, we show how harness engineering (the tools, loops and guardrails around the model) and context engineering (what the model sees at every step) turn accuracy, speed and cost from trade-offs into design choices.
What we will cover
Beyond the Prompt: what harness and context engineering are, how they differ from prompt engineering, and why they become the main levers as agents move into production.
Accurate by Design: grounding agents in structured, governed knowledge and wrapping them in checks and guardrails, so they answer from the right facts instead of guessing.
Fast by Design: designing the agent loop to cut multi-hop retrieval, redundant tool calls and oversized context windows, reducing latency at every step.
Affordable by Design: controlling what enters the context window and how often the harness calls the model, to keep token usage, and inference costs, predictable.
Join us to learn how to build AI agents that are accurate, fast and affordable, without trading one for the other.
⏲️ Agenda
18.30* Welcoming
18.45 Talk
19:30 Q&A
19:45 Closing remarks & Networking and see you at the next Meetup!
*You will receive detailed information on how to access the building close to the date.
🎙️Speaker
Giulio Roggero CTO @ Mia-Platform
25 years of experience in software engineering, serial entrepreneur with more than 10 business initiatives launched, today is co-founder and CTO at Mia-Platform, the Internal Developer Platform named by Gartner Cool Vendor for Software Engineering Technologies and in the Cloud Application Platforms Magic Quadrant.
Principal focus: cloud native, platform engineering, data fabric and omnichannel experience. He likes to paint Blood Bowl miniatures, construct Lego, build and drive RC Cars and learn piano.