PuppyGraph adds AI schema builder and MCP support to help ease enterprise agent adoption

PuppyGraph on Tuesday released PuppyGraph 1.0, adding an AI-assisted graph schema builder and Model Context Protocol (MCP) integrations to its distributed graph query engine, aimed at cutting the complexity, latency and cost of letting AI agents query interconnected enterprise data across thousands of database tables. The schema builder automatically maps existing tables into graph entities, relationships and properties for developer review, while the query engine validates agent-generated queries against the graph schema, rejecting undefined relationships and returning feedback agents can use to correct errors. The release builds on PuppyGraph's 2024 engine, which queries relational databases, data warehouses and lakehouses as a unified Labeled Property Graph without moving or duplicating data; IDC research director Devin Pratt said developers "model entities and relationships once, then query connected data with graph languages instead of writing and maintaining long chains of joins.

PuppyGraph on Tuesday said it was adding an AI-assisted graph schema builder and MCP integrations to PuppyGraph 1.0, the new version of its distributed graph query engine for enterprise data. The update is designed to help developers reduce the complexity, latency, and costs associated with letting AI agents query interconnected data across thousands of database tables and multiple platforms. The AI-assisted schema https://www.infoworld.com/article/2338900/12-schema-design-patterns-for-your-next-app.html builder automatically maps existing tables into graph entities, relationships, and properties, allowing developers to review and approve the resulting schema rather than manually defining how disparate datasets connect, the Santa Clara-headquartered startup said in a statement. These schemas can be used by agents deployed to query enterprise data, giving them a predefined map of how information across different tables and systems is connected, rather than requiring them to identify relevant tables and construct complex SQL joins dynamically, in turn simplifying query generation and reducing errors and retries, it added. To improve query accuracy, PuppyGraph’s query engine also validates agent-generated queries against the graph schema, rejecting undefined relationships and returning feedback that agents can use to correct errors. Further, the execution engine splits complex graph https://www.infoworld.com/article/2265778/what-is-a-graph-database-a-better-way-to-store-connected-data.html queries into smaller operations that run simultaneously across multiple compute nodes, rather than sequentially on a single machine, helping accelerate queries that traverse multiple relationships across large datasets, the company said. These additions build on PuppyGraph’s existing graph query engine, released in 2024, that lets enterprises query relational databases, data warehouses and lakehouses as a unified graph without moving or duplicating data into a separate graph database. The engine maps existing tables into a native Labeled Property Graph LPG model, representing data entities as nodes, their relationships as edges, and their attributes as properties, before executing graph queries across underlying data sources through its own distributed compute cluster. The latest version of the engine, which also supports conventional SQL queries, adds MCP https://www.infoworld.com/article/4029634/what-is-model-context-protocol-how-mcp-bridges-ai-and-external-services.html integrations that allow enterprise teams to connect PuppyGraph to AI agents and tools that support the protocol, the company said. Through these integrations, the supported agents can access PuppyGraph’s graph query capabilities as part of their workflows, helping them query data sources without requiring developers to build custom integrations for each agent, the company said. For developers, the ability to automatically generate graph schemas from existing tables and expose graph-query capabilities to compatible AI agents through MCP can reduce the engineering effort involved in connecting agents to enterprise data, said Devin Pratt https://my.idc.com/getdoc.jsp?containerId=PRF005946 , research director at IDC. “Developers model entities and relationships once, then query connected data with graph languages instead of writing and maintaining long chains of joins. In fact, multi-hop questions, such as who is connected to whom through which accounts, are far more natural in a graph than in SQL, and pipelines to a separate graph store go away,” Pratt added. According to Ashish Chaturvedi https://www.hfsresearch.com/team/ashish-chaturvedi/ , executive research leader at HFS Research, the bigger gain, at least for agentic workflows, is what he terms PuppyGraph’s enforced ontology. The fact that the engine validates agent-generated queries against a predefined ontology, rejecting references to relationships that have not been defined, helps constrain agents to the business entities and relationships established by the enterprise, in turn reducing the scope for agents to generate invalid joins or follow relationships that do not exist in the underlying data, Chaturvedi said. That means agents are less likely to return unreliable results because of faulty joins, Chaturvedi added. That same principle could also help CIOs reduce compute spending on agent queries by limiting the amount of schema information agents need to process and the number of query-generation retries, according to Pratt. Moreover, enterprises can reduce data-management costs and complexity by avoiding duplicate storage and the synchronization pipelines required to maintain a separate graph database, while keeping graph analysis within their existing governed data environment, Pratt said. PuppyGraph’s approach, Pratt added, also aligns with enterprises’ efforts to consolidate data-management capabilities: “IDC research shows that 96% of organizations are actively pursuing such consolidation, while most manage more than 10 database technologies. The high cost of preparing, storing and serving data is also among the most-cited barriers to scaling generative and agentic AI.” However, developers and CIOs will need to weigh several trade-offs before adopting the query engine, analysts pointed out. Developers, for one, will still need to define and maintain the graph model, deciding which tables should become entities, and which keys should define relationships, said Michael Leone https://moorinsightsstrategy.com/team/mike-leone/ , principal analyst at Moor Strategy and Insights. “In other words, the complexity does not disappear; it shifts from writing and maintaining long SQL joins to designing and governing the graph model,” Leone said. Further, the quality of the resulting queries will also depend on the consistency of the underlying schemas and the quality of the source data, echoed Pratt. In addition, Chaturvedi cautioned that performance and costs will depend partly on the underlying data platforms: “Because PuppyGraph queries data in warehouses and lakehouses, running graph-intensive workloads can increase compute consumption on those systems, requiring CIOs to model the costs before scaling deployments.” This also makes workload selection for the query engine important, the analyst pointed out: “The engine is primarily built for analytical queries, rather than applications that require low-latency graph writes or high volumes of transactional updates, which may still be better suited to native graph databases.” CIOs should also consider the maturity and support capacity of PuppyGraph as a smaller vendor before making it a critical part of their data infrastructure, the analyst further said. PuppyGraph said its technology is already being used by enterprises including AMD, Palo Alto Networks, Datadog, Netskope, Coinbase, Trend Micro, Baker Hughes, Sola Security, HighStreet Insurance Partners, and Prevalent AI. The new version of the engine, which supports Apache Iceberg, Databricks, Snowflake, ClickHouse, DuckDB, PostgreSQL, Microsoft Fabric and Apache Hudi among other data platforms, can be deployed using Docker or Kubernetes, or through cloud marketplaces operated by AWS, Microsoft Azure and Google Cloud, according to the company’s installation documentation.