Central Knowledge Graph
XR.Voyage is a cloud-hosted immersive experience platform that aims to gamify exploration, interaction and transformation of your data sets. We operate at intersection of hyperscalers, on-the-fly generated immersive data workstations and knowledge graphs.
As one of earliest adopters of FalkorDB (since early 2019), we chose to base our foundation on best-suited technology core but equally to be surrounded by the support team that is an absolute pleasure to work with.
At Xr.Voyage we run gamification strategy of missions and objectives with a single goal to exponentially increase efficiency of our business partners and the platform itself.
1. We run 100+ LangChain agents to connect and exchange information with 1400+ data providers
2. We maintain two major platform code graphs for python (agentic) and typescript (immersive). We run daily objectives for agents to self-improve
3. All clients share front (immersive) modules as well as back (agentic) sources. However, we run completely isolated back executions per client. We are working at becoming SOC 2 compliant. As part of risk mitigation, we offer FalkorDB Cloud where users can securely self-maintain their own graph FalkorDB instances
4. All the code self-improvement graphs, versioning of code bases, classes, functions are being cross-referenced and mapped with graph node-relationship structures in FalkorDB. Every line of added or modified code is attached to an owner, working out an objective, confirmed with tests
5. In the execution engine, about half of our LangChain agents rely on LLM models. At any given time agents can hot-swap models and LLM providers between OpenAI, Groq, Anthropic, VertexAI, Azure as well as CPU-hosted Ollama models (together on average about 15 models in production)
Our clients usually aim for document stores of processed/transformed datasets and those used in Virtual Reality Simulations. However, versioning of all codes required to assemble a simulation relies on FalkorDB for reference of versions that passed tests. And that can happen even several hundred times per single VR world.
“As one of earliest adopters of FalkorDB (since early 2019), we chose to base our foundation on best-suited technology core but equally to be surrounded by the support team that is an absolute pleasure to work with.”
