AI Engineering in Practice

Real AI lessons, in English, by 8am engineers.

How do you take an AI agent from prototype to production, and what changes for the people who build software around it? At this evening meetup, engineers from 8am share what they have learned building AI into real products: designing MCP tools, rethinking the whole software development lifecycle, and building an agent that works across a law firm’s data without ever showing users something they shouldn’t see. Expect real examples, honest trade-offs and takeaways you can apply next week.

Program
18:00 Arrival & networking
18:10 Welcome
18:15 Building an MCP server – from prototype to production (Mikey Brown)
How 8am took an MCP from an internal prototype to production, and what the team learned about tool design along the way, including how rewriting tool descriptions changed the agent’s behaviour.
18:55 AI transformation – beyond code generation (David Biagi)
Software development has never been just about writing code. How 8am uses AI across the entire SDLC, why the spec becomes the main lever, and where human judgement still matters most.
19:40 What does the agent actually see? Designing an AI agent across a law firm’s data (Gurami Jobava)
How to build an LLM agent that knows how complete its data is, manages a context window that can’t hold everything, and never shows users data they can’t access.
20:15 Panel discussion with 8am engineers
20:35 Wrap-up
20:40 Networking

Each talk is followed by 10 minutes of Q&A. Some software development experience is expected; no prior MCP experience needed. The talks will be recorded.

‍