PuppyGraph raises USD $10 million for AI graph engine

PuppyGraph has raised USD $10 million in a Series A funding round led by Particle Future and defy.vc.

The round brings its total funding to USD $15 million.

The fundraising comes alongside the general availability of PuppyGraph 1.0, a graph query engine designed to let AI agents query enterprise data across multiple systems without moving it into a separate graph database.

The new capital will be used for engineering and enterprise go-to-market expansion. PuppyGraph also said it has seen adoption from companies including AMD, Palo Alto Networks, Datadog, Netskope, Coinbase, Trend Micro, Baker Hughes, Sola Security, HighStreet Insurance Partners and Prevalent AI.

Over the past year, annual recurring revenue grew sevenfold, according to the company.

Agent queries

PuppyGraph is targeting a problem many businesses face as they try to use AI agents on internal data. While a proof of concept may draw on a few dozen tables, production deployments may need to link thousands across identities, permissions, infrastructure and other business systems.

The company argues this becomes difficult as schema descriptions take up more of an agent's context window and the number of possible join paths rises. Copying data into a separate graph database can also create delays and extra maintenance work, it said.

PuppyGraph 1.0 includes an AI-assisted graph schema builder that automatically generates a graph schema from existing tables for review and approval. Users can then investigate connected enterprise data in plain English through PuppyGraph's own AI agent or compatible AI tools connected via MCP.

Its query engine rejects undefined relationship references and returns feedback that agents can use to correct queries. The software can work with data sources including Apache Iceberg, Databricks, Snowflake, ClickHouse, DuckDB, PostgreSQL, Microsoft Fabric and Apache Hudi.

According to the company, the distributed engine supports graph queries at petabyte scale and can complete a 10-hop query across billions of edges in 2.26 seconds on a four-machine cluster.

Security controls

PuppyGraph said its software enforces configured row-level permissions throughout graph traversals, including intermediate steps. The product also includes single sign-on, role-based controls, user impersonation and administrative audit logs.

It added that credential vending allows the system to obtain short-lived credentials when connecting to data platforms such as Snowflake and Databricks, rather than requiring users to expose or manage long-lived credentials.

Weimo Liu, Co-Founder and Chief Executive Officer of PuppyGraph, said the company sees a growing need for agents that can work across a broad range of enterprise data.

"The insights of enterprise data will soon be generated by agents. Users will expect to ask questions that cross departments and systems, beyond the workflows someone built into a dashboard. That requires agents to navigate thousands of tables, follow the right relationships, and retrieve evidence in subseconds," said Liu.

Customer use

Several customers described using the software to avoid maintaining separate graph data pipelines.

"For a user-facing application, query latency and data freshness are non-negotiable. PuppyGraph allows us to execute multi-hop graph traversals directly on our existing data platform without moving or duplicating a single row. By removing the ETL bottleneck, we eliminated pipeline maintenance entirely while delivering lightning-fast graph analytics to our end users in real time," said Ran Xia, Engineering Director at Palo Alto Networks.

Coinbase said the product had changed the work needed to create a graph layer on top of data already stored in Snowflake and Databricks.

"With PuppyGraph, standing up a new graph is a schema definition, not a data migration. We connect directly to the tables we already maintain in Snowflake and Databricks and start querying immediately. No pipelines. No second copy of the data, and no data movement.

The architecture scales with our existing data platform, even at petabyte scale. And with PuppyGraph MCP, we can query complex relationships in real time using natural language. It has become part of how we work with our data every day," said Xinyu Liu, Staff Software Engineer at Coinbase.

Trend Micro highlighted support for Apache Iceberg.

"PuppyGraph is the only solution we found that supports Iceberg natively at that scale, with no data loading required. You just connect to the lakehouse underneath, and the cluster itself doesn't manage any data. Querying it as a graph in place, without duplicating a single row, is what made this viable for us," said Patrick Lu, Lead Architect for Data and AI/ML at Trend Micro.

Sola Security pointed to the scale of its own deployments.

"We run AI agents across thousands of tables and billions of edges a day. Two years ago, I'd have told you the graph side alone made this impractical at this scale. It's one of those products where you ask: how is this just now?" said Leon Goldberg, CTO at Sola Security.

Investors said the appeal lies in helping AI systems execute complex queries on large data sets already held by businesses.

"Big data is still an unsolved problem in the world of AI. You can ask Claude to write a query, but executing it across billions of relationships takes serious engineering. PuppyGraph was built from day one for performance at scale, giving agents a query engine that works directly on the data enterprises already have," said Amy Yin, Partner at defy.vc.