What Leaves With Them: See What a Departure Takes With It
Our own product (RetailBox)
Retailbox R&D: our own product, built for Graph Hacks (FalkorDB and WeMakeDevs), October 2026
An agent that walks a company graph when someone leaves: from the person to their decisions, the systems those decisions constrain, and the owners and teams around them. It then builds a handoff in which every line traces back to its source records.
What was the problem?
When people leave, knowledge leaves with them: systems only they understood, decisions only they wrote down, and the link between two teams that ran through one person. A list of what someone 'owns' misses most of it, and by the time anyone notices, they are gone.
How did Retailbox solve it?
We modelled people, teams, systems, decisions and source records as a graph in FalkorDB. Cypher queries walk up to four steps from the person; a copy of the graph without them shows which systems no one is connected to anymore; a betweenness score finds the people who are the only bridge between teams. The agent then drafts the handoff: who should take each item, and the questions the leaver should answer before their notice period ends. A write gate keeps the memory honest, so no fact enters the graph without a source.
The company and the people in the demo are fictional, generated by a seeded script. The benchmark is small: 8 questions.
What does it look like?
(opens the full-size image in a new tab)
(opens the full-size image in a new tab)What were the results?
- ✓Finds systems and decisions that would be left with no owner, plus lost bridges between teams
- ✓Every handoff line traces back to its source records
- ✓On the same records, plain retrieval answered 0% of the 8 benchmark questions that need the graph
- ✓Live demo, 2:15 video and an open-source MIT repo