SeleneDB FAQs

Question 1

Is SeleneDB suitable for edge computing or constrained environments?

Accepted Answer

Absolutely. SeleneDB is designed edge-first. It's a pure Rust, single ~14 MB binary with zero C/C++ dependencies, offering sub-second cold starts and offline-first sync. This makes it ideal for deployment on constrained devices like Raspberry Pis or factory gateways, as well as cloud environments.

Question 2

What is SeleneDB and its core purpose?

Accepted Answer

SeleneDB is an AI-native graph database built in pure Rust, designed from the ground up for AI workloads. It integrates a full ISO GQL query engine, built-in vector search, graph-native RAG, and AI agent memory into a single, compact binary, optimized for performance and efficiency from edge devices to cloud VMs.

Question 3

How does SeleneDB help reduce token costs for AI agents?

Accepted Answer

SeleneDB significantly reduces AI agent token waste by turning linear context re-establishment costs into constant-time graph lookups. By storing persistent context, prior decisions, and domain knowledge in the graph, it achieves 80-97% token reductions, making AI sessions more efficient and cost-effective.

Question 4

What advanced search and AI features does SeleneDB offer natively?

Accepted Answer

SeleneDB features built-in HNSW vector search, GraphRAG (combining vector similarity, BFS traversal, and community context), semantic search, and agent memory (remember/recall/forget with namespaces). It also supports on-device embedding inference via EmbeddingGemma with GPU acceleration.

Question 5

What query language does SeleneDB use and what are its capabilities?

Accepted Answer

SeleneDB uses ISO GQL (ISO 39075) as its sole query interface. It supports powerful pattern matching, mutations, transactions, variable-length paths, and features a robust 13-rule query optimizer. GQL enables advanced graph algorithms, time-series queries, and direct interaction with AI memory and search features.

SeleneDB

SeleneDB

主な機能

ユースケース

主な機能

ユースケース