Private beta — onboarding data teams now

NerveStax vs dbt Labs — dbt Wizard, dbt Copilot and the dbt MCP server

Agents for everything around your dbt project, on the stack you run.

dbt Wizard is a dbt agent grounded in dbt's own metadata. NerveStax works across dbt Core or platform, Airflow, source loads and on-call, and delivers every change as a reviewed pull request in your repo.

Last verified

NerveStax and dbt Labs across the data lifecycle01 MODEL02 ORCHESTRATE03 TEST04 PROMOTE05 MONITOR06 RESPOND NerveStax and dbt Labs across the data lifecycle01 MODEL02 ORCHESTRATE03 TEST04 PROMOTE05 MONITOR06 RESPOND
NerveStax is the left mark under each stage; dbt Labs is the right. FullPartialNot offered

At a glance

Key facts, side by side.

FactNerveStaxdbt Labs
Focusdbt, Airflow, source loads and on-calldbt development, plus Fivetran ingestion
Where it runsHosted or self-hosted on Kubernetesdbt platform (SaaS); Wizard CLI runs locally
dbt editiondbt Core or dbt platformdbt platform; Wizard CLI also works with dbt Core
OrchestrationYour Airflow, with a verification DAG per changedbt platform jobs and dbt State (preview)
AI usage costYour own LLM keys; private beta is freeWizard metered per token against account credits

The short version

Which one fits your team.

Choose NerveStax if

  • You run dbt Core with Airflow and want agents that change DAGs, schedules and models as reviewed PRs.
  • Your on-call load comes from dbt and Airflow alerts and you want each one triaged and escalated.
  • You want to self-host, use your own LLM keys and avoid per-token platform credits.

Choose dbt Labs if

  • Your team develops in the dbt platform and wants an agent grounded in its native metadata engine.
  • You want column-level lineage, the dbt Semantic Layer and Copilot's one-click SQL, docs and tests.
  • You already load data with Fivetran and want ingestion and transformation from one vendor.

About dbt Labs

What dbt Labs is for.

Fivetran and dbt Labs completed their merger on June 1, 2026. dbt Labs builds dbt and the dbt platform. Its AI tools include dbt Wizard (platform public preview, CLI public beta, Desktop private beta), dbt Copilot for inline generation, and the open-source dbt MCP server. The Fusion engine is open-sourced as dbt Core v2.0 alpha under Apache 2.0.

Your personal dbt agent, available wherever you work.

— dbt Labs, dbt Labs — dbt Wizard product page

Key differences

Where NerveStax and dbt Labs differ.

01

What surrounds the dbt project

NerveStax

NerveStax works on the dbt project and on what runs it: Airflow DAGs, backfills, governed schedules and source loads. Agents share one record of models, tests, runs, schedules and freshness.

dbt Labs

dbt Wizard is built for dbt development: investigating, changing and validating models with lineage, tests, contracts and metrics from dbt's native metadata engine. Orchestration stays in dbt platform jobs.

02

Tested in your own Airflow

NerveStax

Every change builds in an isolated sandbox against a development target, then opens as a pull request with results and reasoning. Airflow changes can also run through a verification DAG in your own Airflow before anyone approves.

dbt Labs

Wizard compiles and builds changes before review and checks upstream and downstream impact. Validation runs through dbt tooling, in the dbt platform or locally with the Wizard CLI.

03

On-call for dbt and Airflow alerts

NerveStax

The on-call agent investigates each alert from dbt, Airflow or any webhook with read-only access, closes noise with a written reason, and escalates real incidents to PagerDuty, Opsgenie or Splunk On-Call. If triage fails, the alert escalates anyway.

dbt Labs

The dbt AI docs list Wizard, Copilot, the Analyst agent and the MCP server; none is described as an alert-triage workflow. The MCP server exposes Admin API tools that an agent you run can use to inspect runs.

04

No platform credits, no lock-in

NerveStax

NerveStax runs hosted or self-hosted with your own LLM keys, on dbt Core or dbt platform, and writes ordinary code to your GitHub or GitLab repo. The private beta is free.

dbt Labs

From September 1, 2026, Wizard usage is metered per token against account credits. Enterprise plans include monthly credits; Developer and Starter plans start with a 30-day trial and $100 in credits.

Feature by feature

The detail, row by row.

Across the lifecycle

StageNerveStaxdbt Labs
ModelFull: dbt changes as reviewed PRsFull: Wizard, Copilot, Fusion engine
OrchestrateFull: Airflow DAGs, governed schedulesPartial: dbt platform jobs, no Airflow
TestFull: Sandbox build + Airflow verificationFull: dbt tests, Wizard validates builds
PromoteFull: PR in your repo; your team mergesFull: Git-backed IDE, review before merge
MonitorPartial: Runs, freshness, drift; no anomaliesFull: Metadata engine: runs, tests, health
RespondFull: Alert triage, escalation to pagingPartial: Run and job tools in MCP server

Capabilities

CapabilityNerveStaxdbt Labs
Works with dbt CoreYesReads the project from your GitHub or GitLab repoPartialWizard CLI supports dbt Core; platform features need an account
Human approval before changes shipYesPull request in your repo; nothing merges without approvalYesReviewable changes with human approval required
Builds and tests before reviewYesIsolated sandbox against a development targetYesCompiles and builds changes, checks upstream and downstream
Airflow DAG changesYesTested by a verification DAG in your own AirflowNoOrchestration is dbt platform jobs
Governed schedulesYesOne tag per model sets cadence; agents never retimePartialScheduled jobs in dbt platform; dbt State in preview
Alert triage and escalationYesdbt, Airflow, webhooks; escalates to PagerDuty, OpsgeniePartialMCP Admin API tools let an agent monitor runs and jobs
Source ingestionPartialMySQL and PostgreSQL, proposed as a pull request (beta)YesFivetran managed connectors (700+)
Column-level lineageNoModel-level lineage todayYesFusion engine; exposed through the dbt MCP server
Semantic layer and metricsNoNot builtYesdbt Semantic Layer
Self-hosted deploymentYesKubernetes with HelmPartialHybrid deployment option on Enterprise+
Bring your own LLM keysYesYour own LLM provider keysYesOpenAI, Anthropic, Bedrock, Gemini, Cortex and more

As of , from each product’s public documentation. See sources below.

In fairness

Where dbt Labs is the better choice.

  1. Wizard reads dbt's native metadata engine: lineage, model health, tests, contracts, run results and semantic definitions. For modeling work inside the dbt platform, that is first-party context from the team that builds dbt.
  2. Fivetran's 700+ managed connectors cover far more sources than NerveStax's beta ingestion, which supports MySQL and PostgreSQL.
  3. Column-level lineage, the dbt Semantic Layer and Copilot's one-click SQL, docs and tests are shipped. NerveStax has model-level lineage and no semantic layer.

Using both

Teams on dbt Core or the dbt platform can use both: dbt Wizard or Copilot for day-to-day modeling, and NerveStax for Airflow DAGs, schedules, source loads and alert triage, with each NerveStax change arriving as a pull request in the same repo.

Questions

Common questions.

Q01Is NerveStax an alternative to dbt Copilot or dbt Wizard?

Partly. Both change dbt projects and ask a person to review. dbt Wizard and Copilot focus on dbt development in the dbt platform or the Wizard CLI. NerveStax also works on the Airflow DAGs, schedules, source loads and alerts around the project. If modeling help inside the dbt platform is what you need, Wizard is a good fit.

Q02Does NerveStax need dbt Cloud or the dbt platform?+

No. NerveStax works with dbt Core or the dbt platform. It reads the project from your GitHub or GitLab repo, builds changes in an isolated sandbox against a development target and opens a pull request. Production runs can stay in your own Airflow.

Q03Can dbt Wizard run on dbt Core?+

Yes, through the Wizard CLI, which is in public beta and supports self-hosted dbt users; a free dbt account gets trial access. Wizard in the dbt platform is in public preview and Wizard Desktop is in private beta. From September 1, 2026, usage is metered per token against account credits.

Q04Does the dbt MCP server do what NerveStax does?+

The dbt MCP server (Apache 2.0) gives any MCP-compatible agent dbt context and actions: CLI commands, Discovery and Admin APIs, the Semantic Layer and SQL. It is a tool layer for an agent you run. NerveStax is a team system: a shared record of runs and schedules, sandboxed builds, pull requests with approval, and alert triage.

Q05What changed with the Fivetran and dbt Labs merger?+

The merger completed on June 1, 2026, with George Fraser as CEO and Tristan Handy as President. It came with dbt Core v2.0 alpha (the Fusion engine under Apache 2.0), dbt State in preview and dbt Wizard. Ingestion and transformation now come from one vendor; NerveStax stays neutral across dbt, Airflow and warehouses.

Q06Does NerveStax replace Fivetran for ingestion?+

Not today. NerveStax ingestion is in beta with MySQL and PostgreSQL sources: each load is proposed as a pull request and schema changes are detected. Fivetran offers 700+ managed connectors. Teams with many SaaS sources should keep a connector vendor.

Early access

See it on the stack you already run.

Beta workspaces open in small batches. Tell us what you run and where the time goes; we reply within a working day. Also worth a look: NerveStax On-Call.

Get early access →