Graph query engine startup PuppyGraph said today it has raised $10 million in new funding to build out its enterprise go-to-market operation.

The round follows the general release of version 1.0 of its software. PuppyGraph provides a query engine that lets companies treat data already sitting in their warehouses and lakehouses as a single graph, without first moving it into a separate graph database. The company says artificial intelligence agents running on it can work across 3,000 enterprise tables and get graph query results back in under a second.

Agent projects tend to hit trouble once they leave the pilot stage, according to PuppyGraph. The company’s own example asks which critical systems an attacker could reach from a compromised account, a question that means tracing how identities and permissions connect across thousands of tables in several systems.

A proof of concept might cover 30 to 50. Teams that scale past that find schema descriptions alone eating into an agent’s context window, and PuppyGraph said queries running several relationships deep can stall until they time out.

Data teams setting up PuppyGraph 1.0 start from a schema that an AI-assisted tool drafts out of their existing tables. Once they have approved it, users can put questions to the graph in plain English.

PuppyGraph has its own agent for that, and the Model Context Protocol lets companies bring in the AI tools they already use. Agents are kept from guessing how the data fits together. Any agent that references a relationship nobody set up gets its query bounced, and the error that comes back is written for the agent to act on.

When PuppyGraph connects to Snowflake Inc. or Databricks Inc. it can obtain short-lived credentials tied to the user’s identity, so nobody has to hand over long-lived keys. Row-level permissions configured on the data still apply, and the company says it checks them at each hop of a traversal.

“The insights of enterprise data will soon be generated by agents,” said Weimo Liu (pictured), co-founder and chief executive of PuppyGraph. “Users will expect to ask questions that cross departments and systems, beyond the workflows someone built into a dashboard.”

PuppyGraph’s customer list includes Advanced Micro Devices Inc. and Palo Alto Networks Inc., and the announcement also named Coinbase Global Inc., Datadog Inc. and Trend Micro Inc. as customers. Xinyu Liu, a staff software engineer at Coinbase, said that because the exchange connects straight to tables it already keeps in Snowflake and Databricks, “standing up a new graph is a schema definition, not a data migration.” Weimo Liu worked as a software engineer at Google LLC before co-founding the Santa Clara-based company.

PuppyGraph’s annual recurring revenue grew sevenfold over the past year, by the startup’s count. No dollar figure was disclosed.

Particle Future and defy.vc led the Series A round. It is the second time defy.vc has led a round for PuppyGraph, after a $5 million seed round in November 2024. PuppyGraph will also put some of the new money into engineering.

Amy Yin, a partner at defy.vc, said writing a query is now a job a user can hand to Anthropic PBC’s Claude. Executing it across billions of relationships still “takes serious engineering,” in her view.

Developers can download a free version of PuppyGraph with a single Docker command and no sign-up form, and the company says it takes about 10 minutes to go from deployment to a first query. Investors have now put $15 million into the startup.

Photo: PuppyGraph