This hackathon is about building with agent harnesses, hands-on, alongside the people building the tools.
We're teaming up with LangChain Ambassador Zamal Babar and Data Reply to host an evening of talks and hacking in Munich. You'll form small groups, build a deep agent from scratch, and demo it to a panel of judges by the end of the night.
What you'll build
The theme is agent harnesses — the scaffolding that turns a model into a system that can plan, use tools, and recover from its own mistakes. Steffen Hausmann will kick things off with two short talks: one on deep agents, and one on managed deep agents. Then you'll put it into practice.
A few days before the event, we'll send out a prompt so you can come with an idea already in mind. Use the time before the hackathon to think through your approach — you'll want it once the clock starts.
What to bring
Your laptop and a group to hack with (we'll help you form one if you don't have one)
Your own API keys for any models or tools you plan to use
Questions for Steffen, Zamal, and the Data Reply team — they'll be around all night
We'll provide LangSmith credits to every attendee for the event, so you can trace and debug your agent as you build.
🏆 Prizes
1st place: $3,000 in LangSmith credits
2nd place: $2,000 in LangSmith credits
3rd place: $1,000 in LangSmith credits
🧑⚖️Judges
Steffen Hausmann, Deployed Engineer at LangChain
Zamal Babar, LangChain Ambassador
A member of the engineering team at Data Reply
⏰ Agenda
5:00 – 5:15 pm Doors open, food, and networking
5:15 – 5:30 pm Welcome and theme intro
5:30 – 5:40 pm Talk: Deep agents (Steffen Hausmann)
5:40 – 5:50 pm Talk: Managed deep agents (Steffen Hausmann)
5:50 – 8:30 pm Hacking (judges circulate and review in the final 30 minutes)
8:30 pm Demos, prizes, and closing
9:00 PM Event Ends
Hosted by Data Reply in Munich. Plentiful LangChain stickers and limited swag for everyone who shows up.
📍Location: Arnulfpark. Address will be sent to approved participants.
About