Link to speaker page
Brian Greene
Brian Greene
Brian
Greene
Company:
Stealth Startup
Fractional Research Leader x AI Builder
LinkedIn profile of
Brian Greene
11:20 am
-
1:20 pm
Theater
Meeting Room 1
Meeting Room 2
Meeting Room 3
talk
workshop
extended talk
panel discussion
dialogue session
Skill level:
sign-up only
From AI Slop to AI Systems (1)
AI makes it easy to produce quick outputs. The challenge is getting it to produce the work you actually want - consistently, without explaining everything again or correcting the same problems every time.System design tackles that problem at its source. Instead of correcting one output at a time, we shape the environment that guides the AI's work.We’ll begin with an honest discussion about where AI is helping us and where it still falls short. Then we’ll put system design into practice. Inside a prepared project directory, we’ll turn context, decisions and ways of working into a system an AI agent can follow, then see how our design choices affect what it produces.You’ll leave with a working system you can keep, use and adapt, plus a practical method for moving from one-off prompting to repeatable, higher-quality work.
Bring your own laptop with Claude Code or Codex installed and signed in — these are the tools Brian demonstrates and supports — plus an active paid plan (from around €20 per month) with enough usage left for a two-hour hands-on session. Your tool needs to be able to open and edit files in a local project folder; running it as an app, from the command line or in an IDE such as VSCode all work, and you should already be comfortable using it. GitHub Copilot or or Gemini CLI can also work, but you'll be running your own setup. Brian will send you a prepared project folder in advance — download it, open it in your tool and complete the short setup check before the session. Please make sure you have a reliable internet connection: you'll get a personal WiFi code for the university network, but some company laptops block it. Test your setup beforehand and keep USB tethering from your phone ready as a fallback.
September 18, 2026 11:20 AM
September 18, 2026 1:20 PM
Europe/Vienna
From AI Slop to AI Systems (1)
AI makes it easy to produce quick outputs. The challenge is getting it to produce the work you actually want - consistently, without explaining everything again or correcting the same problems every time.System design tackles that problem at its source. Instead of correcting one output at a time, we shape the environment that guides the AI's work.We’ll begin with an honest discussion about where AI is helping us and where it still falls short. Then we’ll put system design into practice. Inside a prepared project directory, we’ll turn context, decisions and ways of working into a system an AI agent can follow, then see how our design choices affect what it produces.You’ll leave with a working system you can keep, use and adapt, plus a practical method for moving from one-off prompting to repeatable, higher-quality work.
Bring your own laptop with Claude Code or Codex installed and signed in — these are the tools Brian demonstrates and supports — plus an active paid plan (from around €20 per month) with enough usage left for a two-hour hands-on session. Your tool needs to be able to open and edit files in a local project folder; running it as an app, from the command line or in an IDE such as VSCode all work, and you should already be comfortable using it. GitHub Copilot or or Gemini CLI can also work, but you'll be running your own setup. Brian will send you a prepared project folder in advance — download it, open it in your tool and complete the short setup check before the session. Please make sure you have a reliable internet connection: you'll get a personal WiFi code for the university network, but some company laptops block it. Test your setup beforehand and keep USB tethering from your phone ready as a fallback.
Got some juicy gossip?
We’d love to hear it (and any other questions, wishes or suggestions you have).
