Fivetran and dbt Labs Pitch Open Data Infrastructure as the Base for Agentic Consulting
AI agents are shrinking the hours behind consulting deliverables, and a Fivetran blog post by dbt Labs executive Daniel Bradbury argues that the services opportunity left behind can only be captured on an open data foundation. The post, published on September 28, 2026, calls this approach Open Data Infrastructure (ODI).
- A Fivetran blog by dbt Labs' Daniel Bradbury says AI agents are collapsing the labor behind consulting deliverables, undermining effort-based pricing.
- Citing BCG, the post says up to $200 billion in net-new value pools could emerge over 5 years, and two-thirds of enterprises expect providers to run priority agentic use cases.
- Fivetran and dbt Labs promote Open Data Infrastructure: open table formats, lineage, a semantic layer and governance.
- The post names AI workflow design, AI cost architecture and governed AI deployment as the three service lines.
- Gartner expects more than 40% of agentic AI projects to be abandoned by the end of 2027, according to the post.
Effort-based pricing under pressure
Bradbury, Head of Global Strategic Consulting Partnerships at dbt Labs, opens with a simple comparison. Work that once took 3 analysts 3 weeks, covering sourcing data, reconciling definitions, building a model, validating outputs and assembling a deck, can now be done in an afternoon by an agent connected to properly governed data, he writes.
His argument is that this breaks an industry that has historically priced by the hour or the headcount. The labor needed to produce an analysis has collapsed, while the value of the decision it informs has not. He points to consulting firms' own research as evidence. According to the post, BCG says longstanding delivery models will come under pressure as enterprises adopt systems that can plan and execute entire processes autonomously, and PwC describes the traditional consulting pyramid giving way to an hourglass.
Demand is high, but the product is changing
The post stresses that demand is not falling. It cites Gartner's estimate that more than 60% of organizations intend to deploy AI agents within 2 years, while only 17% have done so today. It also cites BCG research based on 115 enterprise executives and 75 executives at service providers, which estimates as much as $200 billion in net-new value pools over the next 5 years.
Two-thirds of enterprises, according to that research, expect providers rather than their own employees to operate their priority agentic AI use cases. Bradbury's reading is that companies will pay heavily for outcomes but have less reason to pay for the human effort that used to produce them.
Context is the scarce resource
The post argues that consultancies will not be hired to supply intelligence, since frontier AI labs already do that cheaply. The scarce input is context: what the business means, what its systems permit and what an agent is allowed to do. Bradbury notes that Gartner expects more than 40% of agentic AI projects to be abandoned by the end of 2027.
He lists failures he attributes to missing context rather than weak reasoning:
- In 2024, a Canadian tribunal ordered Air Canada to compensate a passenger after its chatbot invented a bereavement-fare policy.
- In July 2025, a Replit coding assistant deleted a production database during a code freeze.
- Klarna, after claiming its AI agent was doing the work of 853 employees and saving $60 million, was later described as a "poster child for bad AI deployments".
He adds that around 80% of enterprises cite data quality as their principal obstacle to AI adoption, and that Fivetran's readiness index found only 15% fully prepared. Settling what counts as a customer, which of 4 competing margin calculations governs decisions, or which churn definition the board uses is judgment work that is hard to automate. If that work sits inside one vendor's proprietary stack, he argues, it cannot be governed end to end or reused for the next client.
What Open Data Infrastructure means
Fivetran and dbt Labs use the term Open Data Infrastructure for an open, governed foundation. At its center is a data lake built on open table formats such as Apache Iceberg or Delta Lake, which let multiple engines read and write the same data safely. Storage is decoupled from compute, so a client can pick the best engine per workload, naming Snowflake, Databricks and GCP.
Three layers sit above it:
- Lineage, which shows where data came from and whether it can be trusted.
- A semantic layer, so humans and agents interpret "revenue" identically.
- Governance, which determines what an agent may see, decide and change.
The post says this gives clients two freedoms: moving between engines without rebuilding definitions, and adopting Databricks Mosaic AI, Snowflake Cortex, Google Vertex AI or Amazon Bedrock, or several at once, because data is not surrendered to any vendor.
Three service lines and a maturity ladder
Bradbury says the $200 billion opportunity can be captured across 3 service lines built on ODI. AI workflow design means modelling data for agent consumption rather than for a dashboard, yielding many small governable agents rather than a monolith. AI cost architecture, which he calls next-generation FinOps, is described as the least contested area because almost no firm has staffed it. Governed AI deployment, he argues, will command the premium: roughly one fifth of enterprises have mature agent governance, and around half of departmental initiatives run with no oversight at all, just as European AI Act obligations begin to apply.
For selling, he proposes placing each client on a ladder, from scattered logic and demonstration agents, through version-controlled transformation and visible costs, to centrally owned definitions in a semantic layer, and finally machine-readable governance. Clients cannot skip rungs, he writes, and a diagnostic is easier to sell than a transformation. The excerpt available ends before the post completes its closing argument.
