Why Autonomous AI Agents Require Zero-ETL Graph Engines Over Naive SQL Queries

Why Autonomous AI Agents Require Zero-ETL Graph Engines Over Naive SQL Queries

Analyst(s): Brad Shimmin
Publication Date: November 16, 2026

PuppyGraph announced a $10 million Series A funding round and the general availability of PuppyGraph 1.0, introducing a zero-ETL graph engine designed to serve as an enforced query layer for autonomous AI agents. By executing distributed multi-hop traversals directly across existing data lakes and warehouses, the platform eliminates the pipeline overhead of legacy graph silos while preventing silent join errors inherent in standard Text-to-SQL workflows. This release echoes a critical industry pivot toward active runtime governance and semantic query enforcement across multi-cloud environments.

What Is Covered in This Article:

  • PuppyGraph 1.0 GA and Series A Funding: Analysis of PuppyGraph’s $10 million Series A financing alongside its reported 7x year-over-year annual recurring revenue (ARR) growth across enterprise deployments.
  • The Limitations of Prompt-Based Schema Stuffing: Architectural evaluation of why packing data dictionaries into model prompts fails at scale, and how an enforced ontology serves as a deterministic runtime firewall.
  • Zero-ETL Distributed Query Mechanics: Breakdown of PuppyGraph’s compute-only engine pushing distributed relational operations down to native lakehouse and warehouse storage layers without data duplication.
  • Competitive Realignment Across Graph and Lakehouse Ecosystems: Strategic assessment of pressures facing dedicated graph databases like Neo4j and TigerGraph, alongside cloud ecosystems like Microsoft Fabric IQ and Databricks Genie.
  • Intermediate Traversal Governance: Real-time enforcement of row-level security and role-based access control across every hop of an autonomous agent investigation.

The News: PuppyGraph has secured a $10 million Series A funding round led by Particle Future, with participation from defy.vc, bringing total capital raised to $15 million. The company reported a 7x year-over-year increase in ARR, propelled by production enterprise implementations across organizations including AMD, Palo Alto Networks, Datadog, Netskope, and Coinbase. The funding coincides with the general availability of PuppyGraph 1.0, transitioning the technology into a hardened data engine tailored for autonomous enterprise agent architectures.

PuppyGraph 1.0 provides an AI-assisted, bottom-up graph schema builder that analyzes existing relational schemas to propose business entities and relationships for data team review and sign-off. The platform features an active ontology validation layer that hard-rejects undefined relationship queries with structured error feedback, enabling agents to self-correct rather than guess at the relationships between database assets. Operating as a decoupled, distributed compute engine, PuppyGraph runs openCypher and Gremlin queries directly across Apache Iceberg, Snowflake, Databricks, PostgreSQL, and ClickHouse environments. The release also incorporates native Model Context Protocol (MCP) support alongside intermediate row-level security enforcement during those multi-hop graph traversals, creating a deterministic governance layer alongside semantic and contextual grounding.

Why Autonomous AI Agents Require Zero-ETL Graph Engines Over Naive SQL Queries

Analyst Take: Even with sizable context windows of one million tokens or more, the established practice of stuffing enterprise schemas into that window via an LLM prompt has hit an operational wall. When teams attempt to scale autonomous agents across thousands of disparate enterprise tables, cramming data dictionaries and foreign key rules into prompt context rapidly exhausts context windows and degrades model reasoning.

More dangerously, tabular databases validate data types rather than business semantics. If an agent writes a syntactically valid SQL query that joins a student identifier to an employee identifier simply because both are integers, the underlying engine executes it without hesitation. This creates a silent join error: a plausible, polished query output that is factually false, which downstream agents subsequently use to compound invalid operational conclusions, all without any apparent error condition.

PuppyGraph 1.0 confronts this vulnerability by turning an approved business ontology into an active execution gatekeeper rather than a passive reference document. Think of prompt context as a paper map that can help an agent reach its goal while also allowing that agent to steer off-road into analytical ditches. Conversely, an enforced ontology functions like a monorail track. If an agent attempts an unmapped relationship traversal, the engine blocks the query and returns structured error as feedback, forcing the agent to inspect the approved graph schema and adapt its plan.

Zero-ETL Execution Squeezes Dedicated Graph Silos

Decoupling graph compute from underlying persistence represents a pragmatic route for enterprise agent adoption. Traditional graph databases require engineering teams to build and babysit complex data pipelines, extracting terabytes of relational data into dedicated graph stores. That operational overhead introduces unacceptable latency, engineering debt, and governance drift.

PuppyGraph hopes to bypass that sort of physical ingestion. Its query engine translates multi-hop traversals directly into optimized push-down SQL queries executed natively by the data warehouse or lakehouse. This promises to deliver subsecond results across billions of edges without moving the underlying records.

PuppyGraph’s architectural approach squeezes legacy graph incumbents. Vendors like Neo4j and TigerGraph have historically relied on native, proprietary storage engines. For this reason, incumbents are likewise looking to deliver frictionless zero-ETL federation across open table formats. Failure to do so would result in their addressable enterprise footprints shrinking into specialized scientific computing domains like the PageRank algorithm and graph neural networks. Concurrently, platform-native tools like Microsoft Fabric IQ and Databricks Genie offer deep context within their respective platforms, but remain architecturally bound to their own environments. PuppyGraph establishes an agnostic layer capable of federating multi-cloud data estates spanning Snowflake, Databricks, and Apache Iceberg.

Capital Rotation Toward the Intelligence Control Plane

The commercial momentum behind PuppyGraph mirrors both a fundamental market reallocation and an architectural re-evaluation. According to the Futurum 2H 2026 Data Intelligence, Analytics & Infrastructure Market Sizing & Five-Year Forecast, the Semantic & Context Layer represents the fastest-growing submarket in data infrastructure, projected to expand at a 26.7% CAGR from $9.3 billion in CY2025 to $38.4 billion in CY2031.

Enterprise capital is shifting away from raw storage tiers toward governed runtime control planes. PuppyGraph is responding to this trend by bringing more deterministic grounding to agents while minimizing critical security exposure by enforcing row-level permissions and role-based access control at each intermediate step of a graph traversal. If an agent pursues an investigative path that crosses into unauthorized data, the engine halts execution mid-traversal. Enforcing runtime query rejection and dynamic security boundaries provides the structural guardrails required to safely deploy autonomous software.

What to Watch:

  • Enterprise Traversal Scope: Monitor whether early adopters expand PuppyGraph beyond initial observability and IT security domains into multi-departmental supply chain and commercial planning workflows.
  • Incumbent Virtualization Responses: Track whether legacy graph vendors introduce native zero-ETL compute engines that query cloud object stores directly, avoiding proprietary database ingestion.
  • Lakehouse Federation Countermoves: Observe whether hyperscalers and lakehouse providers broaden their native semantic layers to execute federated multi-hop traversals across competing cloud platforms.
  • MCP Maturation: Track enterprise production implementations of MCP to determine whether open agent-to-data protocols successfully displace or simply augment custom tool APIs in agentic architectures.

Complete launch details are accessible on the PuppyGraph website.


Disclosure: Futurum is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of Futurum as a whole.

Other Insights From Futurum:

Confluent Positions Streamhouse as the Operational Memory Tier for AI

Oracle AI World: Oracle Introduces Deep Trust AI to Enforce AI Safety at the Database Tier

Solving the Agentic Context Dilemma: Inside Neo4j’s Strategy to Build an Operational World Model

Author Information

Brad Shimmin

Brad Shimmin is Vice President and Practice Lead, Data Intelligence, Analytics, & Infrastructure at Futurum. He provides strategic direction and market analysis to help organizations maximize their investments in data and analytics. Currently, Brad is focused on helping companies establish an AI-first data strategy.

With over 30 years of experience in enterprise IT and emerging technologies, Brad is a distinguished thought leader specializing in data, analytics, artificial intelligence, and enterprise software development. Consulting with Fortune 100 vendors, Brad specializes in industry thought leadership, worldwide market analysis, client development, and strategic advisory services.

Brad earned his Bachelor of Arts from Utah State University, where he graduated Magna Cum Laude. Brad lives in Longmeadow, MA, with his beautiful wife and far too many LEGO sets.

Related Insights
SAS and Posit Bridge the Governed Open-Source Gap in Life Sciences
November 16, 2026

SAS and Posit Bridge the Governed Open-Source Gap in Life Sciences

SAS and Posit PBC announced a strategic partnership integrating SAS Clinical Acceleration with Posit's toolchain, enabling life sciences organizations to streamline governed workflows combining SAS and R code in a...
Frontier AI Tops Out at 45% on Professional Charts
November 16, 2026

Frontier AI Tops Out at 45% on Professional Charts

Frontier AI models score only 45% on professional-grade chart reasoning, revealing a critical capability gap as enterprises deploy AI agents across operations, finance, healthcare, and engineering....
Splunk and Kestra Close the Detection-to-Response Gap
November 13, 2026

Splunk and Kestra Close the Detection-to-Response Gap

Kestra joins Splunk Build Partner Motion with native Add-on enabling enterprise teams to automate remediation workflows directly from Splunk alerts, addressing the persistent detection-to-response gap with governed, auditable processes....
WEKA's First CCO Signals a Shift From AI Pilots to Production Outcomes
November 13, 2026

WEKA’s First CCO Signals a Shift From AI Pilots to Production Outcomes

WEKA names Nic Morales as its first Chief Customer Officer, signaling a strategic shift from AI pilots to production-scale outcomes. The appointment reflects growing enterprise demand for measurable results in...
Motive and Routeware Close the Waste Fleet Data Gap
November 13, 2026

Motive and Routeware Close the Waste Fleet Data Gap

Motive's integration with Routeware eliminates data silos in waste fleet operations, embedding AI dashcam and telematics directly into Routeware's platform for unified visibility and proactive revenue capture....
Dropbox Bets on Agentic AI and Object Storage to Escape Collaboration's Shadow
November 12, 2026

Dropbox Bets on Agentic AI and Object Storage to Escape Collaboration's Shadow

Dropbox executes a two-pronged AI pivot, rolling out agentic capabilities to 70,000+ teams while launching DBX Agent Cloud and acquiring Tigris Data to compete as an independent storage infrastructure layer....

Book a Demo

Welcome

The vision behind everything in Futurum’s Custom Research practice is this: research should show you what is happening, what comes next, and what to do about it. It should be personal to each audience, easy for people to grasp, and structured so LLMs can reason over it accurately. And it should be fast and turnkey; you want answers now, not another project to carry for quarters.

Whether you are defining business, channel, or go-to-market strategy; evaluating vendors or justifying ROI; or commissioning research to fill an emerging market need, we have your back, with a program that answers your questions with the objectivity and credibility to drive real decisions.

To do it, we bring unmatched data to bear: Futurum research, surveys, and market projections; validated market feeds; ETR’s 15 years of insight from 10,000 technology decision-makers; G2’s buyer and user data; and what our analysts hear every day. Add leading primary collection, from AI-moderated voice interviews to surveys and analyst-led interviews, all turnkey, and every project comes out credible, nuanced, and actionable.

And we don’t just drop the results in your lap. For internal work, we provide analyst-led sessions, interactive dashboards, and a range of formats. For market-facing work, Futurum delivers turnkey activation and amplification that actually gets seen, by people and by LLMs, through our media and share of voice. This is research that moves decisions and markets.

We will meet you wherever you are, from a fast-turn brief to a multi-year program, and shape the work to your goals, timeline, and budget. The right program for your moment.

If any of this is useful, I would love to talk.

Benjamin Brown, VP Custom Research, Futurum Research

Benjamin Brown

VP, Custom Research · The Futurum Group

Newsletter Sign-up Form

Get important insights straight to your inbox, receive first looks at eBooks, exclusive event invitations, custom content, and more. We promise not to spam you or sell your name to anyone. You can always unsubscribe at any time.

All fields are required






Thank you, we received your request, a member of our team will be in contact with you.