How dbt Fusion Makes the Modern Data Stack Work Smarter
Published
October 30, 2025
This year at Coalesce, dbt’s annual data conference, one message stood out: data work is moving beyond pipelines and models to something bigger: context and intelligence.
With the launch of the dbt Fusion Engine, dbt introduced a major shift in how developers build and interact with data. It’s not just about faster runs or smarter compilation; it’s about redefining how developers and data teams work with the entire ecosystem. Fusion makes dbt aware of what has changed, where it came from, and how it connects to everything else.
At OneSix, we see this as the start of a new era in analytics engineering. One where the developer experience takes center stage and AI-ready data becomes a real, practical advantage.
A Better Experience for Developers
Developer Experience (DX) is all about how easy and efficient it is to build. In the data world, that means faster testing, easier debugging, and confidence in every deployment.
dbt has always led with this mindset. It gave data teams version-controlled workflows, built-in testing, and clear documentation when few other tools did.
With Fusion, dbt takes DX to the next level:
Stateful performance
The new engine is stateful, allowing dbt to remember what changed since the last run, eliminating unnecessary re-compilation and cutting deployment times by up to 60%.
Cross-platform flexibility
Fusion’s cross-platform compilation allows teams to run the same dbt project across multiple warehouses (Snowflake, BigQuery, Databricks) without rewriting core logic.
Smarter editing tools
The new VS Code extension brings real-time feedback and SQL error detection, inline CTE previews and “compare changes” data diffs, and a smoother, more intuitive development loop.
The result? Building with dbt feels faster, cleaner, and more human than ever before.
Why Fusion is a Big Deal
Behind the scenes, Fusion is more than a performance boost. It’s a full rebuild of how dbt thinks about data. Instead of treating transformations as isolated scripts, Fusion connects everything through a shared layer of metadata. Each model, column, and test becomes part of a living, interconnected system.
This matters because it sets the foundation for the next generation of data tools that can understand what’s happening, not just execute code.
Key breakthroughs include:
Unified metadata architecture
Fusion understands relationships between data assets, making it easier to analyze and automate.
Model Context Protocol (MCP)
An open standard that lets tools and AI agents share lineage, schemas, and dependencies securely.
Think of MCP as a universal translator that allows different systems to “speak the same data language.” For the first time, AI assistants and analytics platforms can work directly with dbt in a governed and meaningful way.
How AI Fits In
The next phase of analytics won’t be about dashboards; it’ll be about understanding. AI thrives on metadata: lineage, ownership, and quality details that describe your data. That’s where dbt’s Fusion Engine and MCP Server come in.
At Coalesce, dbt Labs introduced the remote dbt MCP Server, which allows AI tools like OpenAI, Anthropic, and Cursor to safely connect to dbt projects through a governed API. This enables real, structured collaboration between AI and analytics without sacrificing security or control.
dbt also announced a suite of AI-powered dbt Agents, each designed for specific, safe automation tasks:
- Developer Agent that refactors and validates code.
- Discovery Agent that identifies and recommends trusted datasets.
- Observability Agent that monitors and troubleshoots issues across the stack.
- Governance Agent (in progress) that assists with compliance and ownership management.
Together, these agents represent a shift in how AI supports data teams: not by replacing developers, but by amplifying their capabilities. Instead of guessing, AI can now act on trusted metadata, offering relevant and reliable assistance.
What It Means for Data Teams
For practitioners, Fusion and MCP together will reshape workflows in subtle but powerful ways:
Faster iteration
Incremental compilation and smarter caching allow teams to rebuild only what has changed, speeding up development cycles.
Smarter reviews
Inline previews, data diffs, and emerging AI-assisted analysis make code reviews more visual, accurate, and efficient.
Stronger governance
Column-level lineage and model-aware metadata provide visibility and control across every stage of the data lifecycle.
Cross-platform flexibility
Teams can run models across different warehouses without rework or vendor lock-in, increasing agility and scalability.
Better collaboration
Unified metadata gives analysts and engineers a shared source of truth, improving communication and decision-making.
AI readiness
Structured metadata establishes the foundation for responsible, governed AI integration and intelligent automation.
These improvements reduce friction for individual contributors and create organizational momentum toward more composable, AI-integrated data ecosystems.
The OneSix Perspective
At OneSix, we see dbt Fusion not as a tool update, but as a foundation for the next wave of innovation. Our teams have already begun building Fusion-ready accelerators that connect dbt, Snowflake, and AI-driven metadata platforms. These solutions help organizations evolve from isolated data transformations to intelligent, connected ecosystems.
What excites us most isn’t just what dbt Labs has built; it’s what the community can now build on top of it. With open standards like MCP, the door is open for collaboration across the modern data stack. AI-assisted development, metadata-driven governance, and self-documenting data systems are now within reach.
Get Ready for the AI-Driven Data Era
The message from Coalesce couldn’t be clearer: the future belongs to teams that turn metadata into intelligence. But getting there requires more than adopting new tools requires more than new tools. It takes strategy, architecture, and governance that align around this new way of working.
That’s where OneSix comes in. We help organizations design and implement Fusion-ready, MCP-integrated data ecosystems that connect dbt, Snowflake, and AI. Our goal is to make these advancements actionable—accelerating development, strengthening governance, and unlocking new business insights.
Whether you’re modernizing your dbt setup or exploring metadata-driven AI applications, OneSix can help you turn Fusion’s promise into measurable value. Book a consultation.
Written by
Ozzy Gonzalez, Data Architect