Actian launches VectorAI DB designed for on-premises, edge AI deployments
Actian launched its first vector database with Actian VectorAI DB, which is designed for AI workloads in regulated, disconnected and edge computing environments.
The company, the data and AI division of HCLSoftware, designed its vector database for on-premises deployments. Most vector databases are cloud-first and assume always-on connectivity and when it's not available they don't meet latency and data residency requirements.
Actian VectorAI DB, which was teased by Actian CEO Marc Potter at Constellation Research's recent AI Forum, is designed to run outside of the cloud while keeping sensitive data secure. Actian claims Actian VectorAI DB outperforms open-source vector databases by more than 22x on throughput.
- Actian CEO Potter: AI driving a data governance renaissance
- HCLSoftware’s Actian acquires Wobby, Jaspersoft as it builds out data platform
- Actian adds marketplace, self-serve to Actian Data Intelligence Platform
Emma McGrattan, CTO of Actian, said Actian VectorAI DB is about bringing AI to the data for developers and "deploying vector search wherever data resides so governance, sovereignty, and trust remain built into the architecture."
Key points about Actian VectorAI DB include:
- The vector database doesn't need cloud resources to scale.
- Actian VectorAI DB features low-latency vector search across embedded systems, edge devices ranging from Nvidia Jetson to Raspberry Pi and industrial edge servers, on-premise deployments and cloud.
- Developers leverage the same architecture and APIs in any environments.
- The database embeds multimodal data and supports multiple models.
- Actian VectorAI DB is designed for industries such as manufacturers, healthcare, financial services and government agencies where data sovereignty is required.
- Supported programming languages and SDKs include Python and JavaScript.
- Supported frameworks include LangChain, LlamaIndex and Hugging Face.
- According to Actian, its vector database includes AES-256 encryption, support customer-managed API keys and configurations for various compliance standards.
As for performance, Actian said Actian VectorAI DB delivers 745.2 queries/second on 10M vectors at 768 dimensions.
Actian can fill a key role with its Actian VectorAI DB given that there are enterprises looking to on-prem AI deployments for data sovereignty and control. The Actian VectorAI DB is designed to run in air-gapped facilities, plant floors and production lines for manufacturing, hospital data centers and research facilities, and hybrid environments.
The Actian VectorAI DB is available with a free trial for 30 days for developers. There's also a free Community Edition.