Editorial illustration for PuppyGraph raises USD 10M to help AI agents query thousands of enterprise tables
PuppyGraph Raises $10M for Enterprise AI Query Engine
PuppyGraph has raised $10 million in a Series A round led by Particle Future, with defy.vc also joining in. The round brings the graph query engine startup's total funding to $15 million, money the US-based company says will go toward engineering work and expanding its enterprise sales efforts.
The raise comes right after PuppyGraph pushed its 1.0 version to general availability, a release built around a specific problem: AI agents that need to pull data from enterprise systems often start small, covering maybe 30 to 50 tables, then hit a wall once production demands require linking thousands of tables across different departments and systems. PuppyGraph's platform lets agents query up to 3,000 enterprise tables as a single unified graph, without requiring companies to copy that data into a separate graph database first.
The 1.0 release also adds an AI-assisted schema builder, which generates graph schemas from existing tables for review, plus natural language search through either PuppyGraph's own agent or outside tools linked via the Model Context Protocol. Co-founder and CEO Weimo Liu framed the bet behind the round this way.
“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,” said Ran Xia, Engineering Director at Palo Alto Networks.
Why this matters
The gap PuppyGraph is chasing is a real one: plenty of AI agent demos look great at 30-50 tables and then fall apart when a company tries to point them at the thousands of tables that actually run a business. Weimo Liu's framing, that agents will need to "navigate thousands of tables, follow relationships, and retrieve evidence across departments," is a fair description of where most enterprise AI projects currently stall, not on model quality but on data plumbing. A $10 million Series A led by Particle Future, with defy.vc joining, is a modest bet, not a victory lap, and we'd treat it that way.
The pitch is essentially infrastructure for infrastructure: making existing databases and warehouses legible to agents without forcing a rebuild. For developers and founders, that's worth watching less as a product to adopt today and more as a signal of where the next round of "boring but necessary" AI tooling money is heading. Graph layers on top of relational data have been tried before; whether PuppyGraph's version actually scales past pilot projects is the open question.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives Qantas to Outsource 850 Jobs to India, Citing AI and Automation Shift
From Archives TSMC October Revenue Surpasses August Record With USD 17.81 Bn on AI Chip Demand
Common Questions Answered
What is PuppyGraph's Series A funding round and who led it?
PuppyGraph raised $10 million in a Series A round led by Particle Future, with defy.vc also participating as an investor. This brings the graph query engine startup's total funding to $15 million, which will be allocated toward engineering work and expanding enterprise sales efforts.
What problem does PuppyGraph 1.0 solve for AI agents querying enterprise systems?
PuppyGraph 1.0 addresses the challenge of AI agents needing to pull data from enterprise systems with thousands of tables rather than just dozens. The platform enables agents to navigate thousands of tables, follow relationships, and retrieve evidence across departments without requiring data movement or duplication.
How does PuppyGraph maintain query latency and data freshness according to Palo Alto Networks?
According to Ran Xia, Engineering Director at Palo Alto Networks, PuppyGraph allows execution of multi-hop graph traversals directly on existing data platforms without moving or duplicating data rows. This approach ensures that query latency and data freshness remain non-negotiable requirements for user-facing applications.
Why do most enterprise AI projects currently stall according to the article?
Most enterprise AI projects stall not due to model quality issues, but rather on data plumbing challenges. The gap exists because AI agent demos work well with 30-50 tables but fail when companies attempt to scale them to the thousands of tables that actually run their businesses.
