PuppyGraph, a Santa Clara, California company that makes a graph query engine for data held in warehouses and lakehouses, raised $10m in a Series A round, it announced on 10 November 2026.
Accounts of who led the round differ. PuppyGraph's own blog post says Particle Future led it with participation from defy.vc, while SiliconANGLE reported that Particle Future and defy.vc led it together. defy.vc led the company's $5m seed round in November 2024.
The round brings PuppyGraph's total funding to $15m. No valuation was disclosed.
“The insights of enterprise data will soon be generated by agents.”Weimo Liu, co-founder and CEO, PuppyGraph
What PuppyGraph does
PuppyGraph lets companies treat tables they already store in systems such as Snowflake, Databricks, Apache Iceberg, ClickHouse, DuckDB, PostgreSQL and Microsoft Fabric as a single graph. Users can then follow relationships across many tables without first copying the data into a separate graph database.
The company pitches this at AI agents that need to answer questions spanning many departments and systems. It says its engine already lets agents query 3,000 enterprise tables as one graph. Users can ask questions in plain English through PuppyGraph's own agent or through other AI tools connected over the Model Context Protocol.
The funding followed the general release of PuppyGraph 1.0. That version adds a tool that uses AI to draft a graph schema from existing tables for a team to approve, and checks that reject queries referring to relationships the schema does not define, returning errors an agent can act on. It also runs multi-hop queries across a cluster, applies row-level permissions at each step of a traversal and, with Snowflake or Databricks, uses short-lived credentials tied to the user, according to SiliconANGLE.
The company reported a benchmark in which a 10-hop query across billions of edges finished in 2.26 seconds on four machines, and noted that results depend on the workload. A free version runs in Docker without sign-up, and the company says a first query takes about 10 minutes.
Customers and growth
Companies using PuppyGraph include AMD, Palo Alto Networks, Datadog, Netskope, Coinbase, Trend Micro, Baker Hughes, Sola Security, HighStreet Insurance Partners and Prevalent AI, according to the company.
PuppyGraph said its annual recurring revenue grew sevenfold over the past year but did not give a dollar figure.
Xinyu Liu, a staff software engineer at Coinbase, said, according to SiliconANGLE, that with PuppyGraph setting up a new graph means defining a schema rather than migrating data. Amy Yin, a partner at defy.vc, said in comments reported by SiliconANGLE that AI tools such as Claude can now write queries, but running them across billions of relationships still takes serious engineering.
Who runs it and what the money is for
PuppyGraph's co-founder and chief executive, Weimo Liu, was previously a software engineer at Google, SiliconANGLE reported.
The company said the Series A will pay for engineering and for expanding its enterprise sales and go-to-market work. SiliconANGLE reported that most of the money is aimed at go-to-market, with some going to engineering. PuppyGraph said its priorities include cutting the work needed to build graph models and making relationship queries practical at enterprise scale.
Company website: puppygraph.com
