For AI apps anytime & anyplace
AI on any device even without internet
✔ On-premise, on-device
✔ No cloud dependency
✔ Uninterrupted experiences
Keep your AI app in sync
Out-of-the-box seamless synchronization
✔ Sync on the edge & to the cloud
✔ Host Data Sync on any device
✔ 100% transactionally safe
So fast, you forget its tech
Efficient resource-use for more speed
✔ Mobile phones, IoT, wearables
✔ The kind of fast your customers feel
✔ Ultrafast on constrained devices
On-device Vector Search for Embedded, Mobile & IoT devices
| 1 2 3 4 5 6 7 8 | objectbox = ObjectBox( embedding=HuggingFaceEmbeddings(), embedding_dimensions=768, distance_type=VectorDistanceType.EUCLIDEAN, db_directory="obx-db" ) result = objectbox.similarity_search("What is ObjectBox?", k=1) print(result) |
| 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 | @Entity(id=1, uid=101) # Note: id, uid, and model setup are only required in the alpha version. # The final version will handle this automatically! class Document: id = Id(id=1, uid=1001) title = Property(str, PropertyType.string, id=2, uid=1002) content = Property(str, PropertyType.string, id=3, uid=1003) author = Property(str, PropertyType.string, id=4, uid=1004) year = Property(int, PropertyType.int, id=5, uid=1005) embedding = Property(list, PropertyType.floatVector, id=6, uid=1006, index=HnswIndex(id=1, uid=10001, dimensions=1024)) model = Model() model.entity(Document, last_property_id=IdUid(6, 1006)) model.last_entity_id=IdUid(1, 101) model.last_index_id=IdUid(1, 10001) store = Store(model=model, directory="obx-db") |
| 1 2 3 4 5 6 7 8 9 10 11 12 13 | @Entity() class City { @Id() int id = 0; String? name; @HnswIndex(dimensions: 2) @Property(type: PropertyType.floatVector) List<double>? location; City(this.name, this.location); } |
A few lines of code speak louder than a thousand words
Easily get started within seconds
With just a few lines of code, start using vector search to find the relevant data for RAG, detection, search, recommendation or any other AI-supported task; no cloud needed
Enjoy the speed while you scale while we take care that you can launch and expand your applications with minimal work
Works on any POSIX system, e.g.
Tech Highlights
Advanced ANN Search AlgorithmOur Hierarchical Navigable Small World (HNSW) algorithm delivers industry-leading performance and scalability for precise and fast responses.
Ground your AI app in your data using Retrieval Augmented Generation (RAG) for personalized, enhanced, and up-to-date LLM responses.
Superfast Semantic & Hybrid SearchCombine vector searches with other query conditions, creating a flexible and powerful search capability that includes non-vector data and can link to several objects.
Because no one likes waiting. 10X faster than any alternative paired with an incredebly lightweight footprint.
Disk-Based & RAM Caching Speed
ObjectBox runs efficiently on constrained devices, bringing robust performance to widely accessible hardware.
Due to its efficiency, ObjectBox reduces resource-use (CPU, Memory, energy…) and therefore CO2, time, and money waste
For the low-latency “always-on” experience. Develop applications that work on- and offline, unburdened by the need for a constant Internet connection.
Data ControlSelf-host, deploy locally, or run on-premise – ensure data sovereignty, compliance, and seamless performance even in low-connectivity environments.
Never miss a beat. Our offline Data Sync keeps data flowing seamlessly across devices on the edge of the network and to the cloud. 100% cloud optional.
