What matters in AI.

Subscribe

PuppyGraph gets $10 million to control how AI agents query data

PuppyGraph got $10 million and says its 1.0 engine stops agent queries that break an approved map of business data.

PuppyGraph makes a graph engine, a tool that finds connections between data items. It got a Series A investment of $10 million, led by Particle Future. The total is $15 million. The tool is made for AI agents that query company data. The company released version 1.0 and says its yearly revenue is 7 times that of the year before.

How it works

The engine runs on top of data stores such as Snowflake, Databricks and Apache Iceberg. It does not copy the data. It stops an agent query that uses a connection that is not on the approved map. The agent gets an error and can try again.

The proof

The company says that AMD, Datadog and Coinbase use it. It gave no test results. An analyst says the engine promises results in less than 1 second across billions of connections.

What is not known

The analyst says the market for semantic and context layers can become $38.4 billion in 2031, from $9.3 billion in 2025. That is an estimate, not a result. The analyst also says Neo4j and TigerGraph must give the same access with no copy.

Insight: Three suppliers sell the layer that sets business meaning

Neo4j released Context Engine on 20 October 2026, with no test results. Databricks released Genie Ontology on 11 November 2026. PuppyGraph adds a neutral layer that stops bad queries. This seems to be a race to be the owner of the definitions that agents use. The first layer that a company selects can set the rules. Select the owner of your definitions before you select a supplier.

Sources

  1. Why Autonomous AI Agents Require Zero-ETL Graph Engines Over Naive SQL QueriesThe Futurum Group
AI MATTER · NEWS · AI MATTER · NEWS ·16 NOV2026

Posted

Learn the terms in this story

More in Deals

All Deals news