PuppyGraph raised a $10 million Series A to expand engineering and enterprise sales for software that lets AI agents investigate company data without copying it into a separate graph database. Particle Future and returning investor defy.vc backed the round, announced November 10.
In a company-published case study, AMD had already gathered enterprise records in a shared data platform, but AI models still struggled to work out how to join information across many tables. PuppyGraph added explicit links between records without requiring AMD to maintain a separate graph database. AMD kept its existing data-loading pipelines.
The financing follows the general release of version 1.0, which includes an AI-assisted tool that proposes links between records for a data team to approve. The engine rejects queries that refer to links outside that model. PuppyGraph says this guards against invented connections, while acknowledging that an agent can still reach a wrong conclusion.
The Series A is twice PuppyGraph’s $5 million seed round, announced in November 2024 and led by defy.vc. The new financing brings total funding to $15 million.
Annual recurring revenue rose sevenfold over the past year, according to PuppyGraph, which did not disclose the dollar total. Enterprise pricing is based on the memory and processing capacity of the servers running its software, with no per-query or storage charges.