Vespa Integrations

Coral is a knowledge assistant for enterprises to supercharge the productivity of their most strategic teams. Engage Coral with a prompt to find answers from across your documents backed with citations. Generated responses are verifiable with citations to sources, mitigating against hallucinations. Explain large language models to a non-technical retail exec. Whether for finance, support, sales, or other teams, tailor Coral to your unique job functions. Make it more powerful by connecting data sources that augment its knowledge base. Coral connects to your ecosystem with 100+ integrations across CRMs, collaboration tools, databases, and more. Manage Coral within your own secure cloud, whether through cloud partners (AWS, GCP, OCI, etc.) or virtual private clouds. Coral data is never sent to Cohere, it stays within your own environment. Coral's responses can be grounded on your data and documents. Users will see citations to denote where responses are based from.

Put your data to work, wherever it resides, with the open, hybrid data lakehouse for AI and analytics. Connect your data from anywhere, in any format, and access through a single point of entry with a shared metadata layer. Optimize workloads for price and performance by pairing the right workloads with the right query engine. Embed natural-language semantic search without the need for SQL, so you can unlock generative AI insights faster. Manage and prepare trusted data to improve the relevance and precision of your AI applications. Use all your data, everywhere. With the speed of a data warehouse, the flexibility of a data lake, and special features to support AI, watsonx.data can help you scale AI and analytics across your business. Choose the right engines for your workloads. Flexibly manage cost, performance, and capability with access to multiple open engines including Presto, Presto C++, Spark Milvus, and more.

​Kong AI Gateway is a semantic AI gateway designed to run and secure Large Language Model (LLM) traffic, enabling faster adoption of Generative AI (GenAI) through new semantic AI plugins for Kong Gateway. It allows users to easily integrate, secure, and monitor popular LLMs. The gateway enhances AI requests with semantic caching and security features, introducing advanced prompt engineering for compliance and governance. Developers can power existing AI applications written using SDKs or AI frameworks by simply changing one line of code, simplifying migration. Kong AI Gateway also offers no-code AI integrations, allowing users to transform, enrich, and augment API responses without writing code, using declarative configuration. It implements advanced prompt security by determining allowed behaviors and enables the creation of better prompts with AI templates compatible with the OpenAI interface.

AccessOwl is an Access Governance and SaaS management tool that simplifies the management of employee access to SaaS applications, from onboarding to offboarding. As the central hub for SaaS access, it eliminates the need to ask who manages specific tools or what approval is required, while tracking every app, user access, and permission used in the organization. AccessOwl automates user account provisioning, access requests, approvals, reviews, and Shadow IT detection, helping teams replace spreadsheets with a source of truth and reduce the risk of missed offboardings. It integrates with Slack so employees can request access directly where they already work, and HRIS integrations automate employee onboarding and offboarding while syncing employee details such as title, department, and manager. AccessOwl can provision and deprovision users across hundreds of SaaS applications without requiring SCIM or SAML.

Voyage AI introduces voyage-code-3, a next-generation embedding model optimized for code retrieval. It outperforms OpenAI-v3-large and CodeSage-large by an average of 13.80% and 16.81% on a suite of 32 code retrieval datasets, respectively. It supports embeddings of 2048, 1024, 512, and 256 dimensions and offers multiple embedding quantization options, including float (32-bit), int8 (8-bit signed integer), uint8 (8-bit unsigned integer), binary (bit-packed int8), and ubinary (bit-packed uint8). With a 32 K-token context length, it surpasses OpenAI's 8K and CodeSage Large's 1K context lengths. Voyage-code-3 employs Matryoshka learning to create embeddings with a nested family of various lengths within a single vector. This allows users to vectorize documents into a 2048-dimensional vector and later use shorter versions (e.g., 256, 512, or 1024 dimensions) without re-invoking the embedding model.