VisualDBT — the desktop platform for dbt + Snowflake + Fivetran teams. THE PROBLEM Modern data teams use dbt for transformation, Fivetran for ingestion, and Snowflake for warehousing — but no single tool shows the full pipeline end to end. When something breaks, impact analysis across tool boundaries is manual. When you ship a new pipeline, you rewrite staging-and-mart boilerplate over and over. Cloud lineage SaaS alternatives cost $25,000–$50,000/year, send your data to vendor cloud, and lock you into one vendor's UX. VisualDBT is a desktop app that does three things in one place: see your pipeline, observe what's happening in it, and (new in v1.4) build new pipelines that match your team's existing code. VISUALIZE • Interactive DAG of your full pipeline — source databases → Fivetran → dbt → Snowflake → Looker • True column-level lineage parsed from compiled dbt SQL (CTEs, JOINs, subqueries, window functions, UNIONs) with confidence scoring • Impact analysis: simulate column removal, model deletion; see every downstream asset that breaks • 15+ layer filters: sources, seeds, snapshots, staging, intermediate, marts, exposures, semantic models, metrics, BI dashboards • Multi-project dbt Mesh with cross-project ref resolution • Heatmap overlay: green-to-red rings on every node by cost, query count, or row count OBSERVE • Snowflake cost attribution per dbt model from QUERY_HISTORY (top spenders, per-model trends, click-to-graph) • Run history page — every dbt run persisted, per-model gantt timeline, bar width = execution time • Execution-time anomaly detection — flags any model running 2× or more slower than its rolling-10-run median • Row-count anomaly detection with rolling z-score (≥ 2.5σ warning, ≥ 4σ critical) • Schema drift between syncs with rename detection • PII auto-tagger — 10 classifications (email, phone, SSN, credit card, addresses, DOB, name, IP, MRN, tax ID) • Live row preview — right-click any model, see 100 rows from Snowflake with PII auto-masked • Tests as graph elements with one-click run • OpenLineage 2.0 emitter — plug into Marquez, Datadog, Astro, Atlan, or any compatible backend • Fivetran connector control — sync now / pause / resume from the graph; live sync-health badges • Fivetran ↔ dbt schema drift — find synced-but-unused columns, missing-but-referenced columns, dead tables. The unique seam neither Fivetran's UI nor dbt Cloud can analyze alone BUILD (NEW IN v1.4) • 5-step wizard scaffolds new dbt pipelines that mimic your project's existing conventions • Archetype miner walks your graph and extracts naming, materialization, test, and folder conventions from real models • Live source schema introspection from Fivetran connectors, Snowflake INFORMATION_SCHEMA, or dbt sources.yml • PII columns auto-flagged inline before any code is generated • Three authoring paths in one wizard: – One-shot Generate: deterministic templates + optional AI refinement per file – AI Pair Chat: multi-turn pair-programming with file proposals you Accept / Reject / Edit inline – Context Pack export: download a self-contained Markdown prompt for use with Claude Code, Cursor, Augment, Cline, Aider, or Continue • Multi-turn + native tool-use AI on Anthropic, OpenAI, Gemini, Ollama • Sandboxed Apply: writes only inside your dbt project root, refuses to overwrite without confirm, creates a fresh git branch, runs dbt parse for verification before you open a PR INTEGRATIONS • dbt Core + dbt Cloud (manifest.json, catalog.json, run_results.json) • Snowflake (INFORMATION_SCHEMA, QUERY_HISTORY, ACCOUNT_USAGE) • Fivetran (connectors, schemas, sync control via REST API) • Looker (explores, looks, dashboards) • 7 AI providers: Anthropic Claude, OpenAI, Google Gemini, Ollama (local + cloud), Claude CLI, Augment CLI • Slack, Microsoft Teams, generic webhook outputs • OpenLineage 2.0 emit (Marquez, Datadog, Astro, Atlan) • MCP Server for Claude Code, Cursor, Windsurf • Plugin system for custom connectors PRIVACY-FIRST • Desktop application — works offline • All data stays on your machine • Credentials encrypted with OS keychain (Windows DPAPI) • No cloud dependency, no telemetry, no data leaving your device • AI providers: bring your own API key. We never see your prompts. • OpenLineage emit is opt-in and only goes to your configured endpoint PRICING • FREE — full core: interactive DAG, impact analysis, all integrations, PNG/CSV/JSON exports • PRO $19/mo — cost attribution, column lineage, run history, AI features, the new Build feature • TEAM $39/mo — PII auto-tagger, GDPR cascade, plugins, dbt Mesh, MCP server • 30-day in-app trial of every feature — no credit card required REPLACES • Cloud lineage SaaS at $25K–$50K/year • DataHub / OpenLineage cloud setups that need Kubernetes • dbt Cloud's lineage view at $100/seat/month • Manual cost-attribution spreadsheets • Hand-rolled scaffolding scripts for every new pipeline Built by data engineers, for data engineers. One-click install from the Microsoft Store. No Docker, no Kubernetes, no setup.