StarTree on Wednesday unveiled new AI-powered capabilities, including support for Model Context Protocol, aimed at better enabling users to get real-time insights from AI applications.
Model Context Protocol (MCP) is an open standard developed by AI startup Anthropic. It sets guidelines for how large language models (LLMs) interact with databases and other data sources and is designed to help AI agents perform real-world actions.
StarTree will launch support for MCP in June, enabling LLMs such as ChatGPT and Google Gemini to securely access data in StarTree so that agentic AI applications developed with LLMs possess the contextual awareness to meet an individual enterprise’s real-time needs.
In addition, StarTree plans to make Vector Auto Embedding generally available in the fall. The feature speeds and simplifies generating vector embeddings that enable access to unstructured data, power similarity searches and feed the retrieval-augmented generation (RAG) pipelines that help train models.
Given that MCP has emerged as a key enabler of agentic AI by providing connectivity between agents and the data needed to feed them, StarTree’s addition of support for the open standard is significant for the vendor’s users, according to Matt Aslett, an analyst at ISG Software Research.
“Support for MCP will be critical in enabling applications developed and deployed on StarTree to participate in and benefit from agentic workflows,” he said. “[It will] facilitate support for real-time intelligent applications that provide interactive and personalized interfaces and AI-driven recommendations.”
Based in Mountain View, Calif., StarTree is a database-as-a-service vendor whose platform is built on the open source Apache Pinot online analytical processing database framework. To date, the 2019 startup, which specializes in fueling real-time analytics workflows, has raised $74 million in venture capital funding, including $47 million in 2022.
New capabilities
Generative AI (GenAI) has the potential to transform enterprises, making workers better informed as well as more efficient. Consequently, many enterprises have increased their investments in AI development since OpenAI’s November 2022 launch of ChatGPT marked a significant improvement in GenAI technology.
Generative AI alone, however, isn’t transformational for enterprises. LLMs and other GenAI models need to be trained on an organization’s proprietary data to understand its operations and benefit its workers.
Training models to understand the unique characteristics of enterprises, meanwhile, is complex, requiring the development of an intricate pipeline that ensures data quality, retrieval of only relevant data, and the security of data as it’s combined with other systems. In addition, it often includes transforming unstructured data to make it discoverable.
Many data management vendors have responded to the surging interest in AI development by providing customers with tools that simplify AI pipeline development and model training. For example, Databricks, Snowflake, MongoDB and Couchbase are among the many data storage providers now offering tools that simplify AI development.
StarTree already provides vector indexing capabilities and other tools to help customers develop AI applications. MCP support and Vector Auto Embedding aim to improve on those existing capabilities to provide the speed and real-time data that agents require to surface relevant insights and autonomously perform certain tasks.
Through the additions of MCP support and Vector Auto Embedding, StarTree will enable the following:
- AI agents to analyze live, structured enterprise data in real time via MCP.
- Natural language querying, which deploys MCP to simplify natural language to SQL translations and makes them easier to deploy.
- Real-time RAG that feeds a continuous flow of data from its source through the embedding process to agentic AI applications used for purposes such as system observability.
Just as support for MCP will be significant for StarTree users, so will Vector Auto Embedding, according to Aslett.
“Automating the generation of vector embeddings has the potential to simplify and accelerate RAG, which is likely to be important for StarTree users given that the database is typically deployed to support data-intensive operational applications that rely on real-time analytic processing,” he said.
In addition to adding support for MCP and developing Vector Auto Embedding, StarTree on Wednesday launched the general availability of Bring Your Own Kubernetes (BYOK). BYOK is a new deployment option that lets organizations take full control over StarTree within Kubernetes environments, irrespective of whether they’re in the cloud, on-premises, or part of a hybrid architecture.
Motivation for adding the new capabilities stemmed from a combination of StarTree’s observation of market trends and customer feedback, according to Chinmay Soman, the vendor’s head of product.
And while all aim to address user needs, MCP support stands out as potentially most important, given that it provides a simple way for agents to interact with StarTree’s real-time analytics database, he continued.
“We already had the concurrency and scale piece solved,” Soman said. “What was missing was an easy, robust way for agents to interact with a real-time analytics database. MCP fills that gap.”
Looking ahead
With MCP support and Vector Auto Embedding now in preview and BYOK generally available, StarTree is focused on making its platform easier to use, according to Soman.
In addition, the vendor plans to continue adding new capabilities such as improving the scale that Pinot can handle and isolating workloads.
Aslett, meanwhile, suggested that database specialists focused on enabling real-time analytics such as StarTree, CelerData, ClickHouse and Imply all need to do more to raise their profiles.
Many enterprises aren’t familiar with specialized vendors like StarTree and its competitors because tech giants such as AWS, Google Cloud, Microsoft and Oracle all provide a variety of databases, and longtime database specialists, including MongoDB and Couchbase have had time to gain recognition.
“They face an uphill battle to raise their profile, given the dominance of the market by established database and cloud providers,” Aslett said. “Highlighting innovative customers taking advantage of differentiating functionality will go a long way to raising their profile.”
Source: https://www.techtarget.com/
