Graphs Gone Wild - Mini Hack
East Berlin — Germany

When
10/23/2026, 1:00:00 PM
Where
Donaustraße 44, 12043 Berlin, Germany, Berlin, BE, Berlin, BE, de
About
Graphs Gone Wild is for people experimenting with AI agents, memory, documents, graphs, MCP, and new developer tooling.
Use whatever stack you like: Cursor, Claude Code, Codex, Copilot, your own agent setup, frameworks, databases, models, or something completely experimental.
For this NODES 2026 pop-up, Neo4j will provide some of the building blocks, including Document Intelligence, Aura MCP, knowledge graphs, and Agent Memory.
More tools, technologies, and partners will be revealed soon.
The goal: build something interesting in a few hours. It does not need to be polished. A working experiment and a good demo are enough.
What can you build?
Anything around agents, connected knowledge, memory, documents, tools, or graphs.
Some ideas:
Turn documents into knowledge an agent can explore
Give an agent persistent or contextual memory
Build an agent that reasons across connected data
Combine RAG with knowledge graphs
Connect agents to tools through MCP
Experiment with multi-agent workflows
Bring your own dataset and graph it
Build something weird that nobody asked for
Things to play with
Neo4j Aura for connected data and knowledge graphs
Neo4j Document Intelligence for turning documents into structured knowledge
Aura MCP for connecting agents and AI applications to Neo4j
Neo4j Agent Memory for persistent memory and connected context
More partners and tools to be revealed
These are starting points, not requirements.
What should you bring?
Your laptop
Your preferred coding agent or AI development environment
API keys for the models and services you use
An idea, dataset, or documents if you have something to explore
Or just show up and find something to build.
Who is this for?
AI engineers, developers, founders, researchers, product builders, designers, and anyone experimenting with coding agents.
You do not need Neo4j experience. You should be comfortable experimenting with code, APIs, and AI tools.
What counts as success?
A small prototype. A we