PuppyGraph Raises $10 Million to Scale Enterprise Graph Queries

By Nathan Brooks

PuppyGraph announced a $10 million Series A on November 10, 2026, led by Particle Future with participation from defy.vc. The company said the financing brought total funding to $15 million and would expand engineering and enterprise go-to-market work.

The round followed the general availability of PuppyGraph 1.0. PuppyGraph reported sevenfold annual recurring revenue growth over the previous year, citing AMD, Palo Alto Networks, Coinbase, Trend Micro, Datadog and Sola Security as customers.

A graph engine for agents working in place

PuppyGraph develops a graph query engine for AI agents that investigate business questions across departments, rather than merely generating SQL for a known dataset. An agent may need to connect customers, products, components and suppliers; as tables multiply, schema descriptions consume more of its context window while possible join paths increase. The engine applies graph computation directly to supported databases, warehouses and data lakes.

Records remain in systems such as Apache Iceberg, Databricks, Snowflake, PostgreSQL and Microsoft Fabric, according to PuppyGraph. That design avoids a dedicated graph database, a separate graph copy and the associated loading pipelines. Applications can still use SQL for filters and aggregations while choosing graph queries for relationship-heavy questions.

An ontology supplies approved business entities and relationships, giving agents valid investigation paths. PuppyGraph’s AI-assisted schema builder proposes that model from existing tables, while data teams review and approve its business meaning. Compatible AI tools connect through MCP, and users can also work through PuppyGraph’s built-in agent.

The engine rejects references to entities or relationships outside the ontology and returns structured feedback for query correction. PuppyGraph says this prevents silent errors when an agent invents a connection and treats it as evidence. The company cautions that validation does not guarantee a correct conclusion, because an agent can still pursue a poor hypothesis or conduct an incomplete analysis.

Speed that can support iterative investigations

PuppyGraph reported that its engine supports subsecond graph queries for supported workloads and lets agents query across 3,000 enterprise tables as one graph. In a separate benchmark, PuppyGraph said a four-machine cluster completed a 10-hop query across billions of edges in 2.26 seconds, with performance dependent on the workload.

The company said speed matters because agents test multiple hypotheses and a slow query can stall the next dependent analytical step. PuppyGraph said the benchmark therefore bears on whether deep relationship searches can remain interactive, rather than serving only as a measure of raw query throughput.

Controls for identity and data access

PuppyGraph 1.0 enforces configured row-level permissions throughout graph traversals, including intermediate steps. Role-based controls and user impersonation tie access to user identity; the engine also supports single sign-on and administrative audit logs that record management activity. Credential vending obtains short-lived, identity-aware credentials for Snowflake and Databricks without requiring users to expose or manage long-lived credentials.

Customers are extending deployments across departments

Sola Security CTO Leon Goldberg said: “We run AI agents across thousands of tables and billions of edges a day.” Palo Alto Networks engineering director Ran Xia said the company runs multi-hop traversals on its existing data platform without moving or duplicating rows, eliminating the associated ETL pipeline maintenance.

Coinbase senior staff engineer Xinyu Liu said Coinbase connects directly to tables maintained in Snowflake and Databricks and uses MCP for natural-language relationship queries. AMD began with IT workloads and, according to PuppyGraph, is expanding toward information-security and digital-marketing data. PuppyGraph said the expansion illustrates its broader model: as a relationship map connects more departments, agents can investigate questions that previously stopped at organizational boundaries.

What does the funding change for enterprise data teams?

PuppyGraph plans to use the capital to reduce graph-model setup work, make relationship queries practical at enterprise scale and help customers move agents into production. In practical terms, the change is a lower implementation burden for teams that need relationship-heavy analysis across existing data platforms. Human oversight remains part of the model: data teams review and approve ontology meanings, while configured permissions govern what each user can see.

The central challenge, as PuppyGraph frames it, is to broaden investigations without weakening human-approved ontologies or permission controls. The new funding supports expansion in engineering and enterprise go-to-market work, but those controls still determine which relationships are valid and which data each user can see. For buyers, the test is whether agents can investigate deeply across departments while keeping query paths, identity and source data under control.