Make more
of your data.

CSV and Excel files, relational databases, NoSQL sources and HTTP APIs. Bring data in once or make the load repeatable.

• Embedded DuckDB; Postgres, MySQL/MariaDB, SQL Server/Azure SQL, Redshift, SQLite and DuckDB-file connections
• MongoDB, Cassandra and Firestore snapshot imports
• No-code API builder with authentication, pagination and response previews
• Replace, append or upsert loads; schema-drift controls, cancellation and run logs

• SQL and Python transformations with notebook cells and Markdown notes
• No-code filtering, joins, aggregation and calculated columns with step-by-step previews
• Visual pipeline graphs, nested pipelines and if/else gates
• Success, failure and completion dependencies; cancellation and recorded run outcomes
• CLI scheduling and background operation while the app window is closed

Python is installed separately. Scheduled and background work needs the host machine and QDP runtime running.

• Semantic models with entities, relationships, measures, metrics and drill hierarchies
• Multiple output ports with independently versioned Parquet contracts
• Schema comparisons, content hashes and quality-gated publishing
• A consumer catalogue with schemas, previews, quality results and version history
• DPROD JSON-LD, native table materialisations and generated dbt model files and semantic YAML
• Meaning manifests, shared dimensions and an ontology of objects, links and governed actions

QDP writes dbt artefacts into your project; dbt and MetricFlow are external tools and are not run by QDP.

• Governance Explorer with ownership, certification, tags and asset posture
• Quality checks, freshness expectations and a history of results
• Table and column lineage, audit history and workspace usage insights
• Rule-based classification proposals for sensitive data, with human review
• Identity-aware access policies, column masks and row filters for supported agent and sharing paths

The local workspace owner remains in control. Policies do not create a security boundary against that owner.

• SQL Query with autocomplete and read-only exploratory notebooks
• SQL, Python, Markdown and chart cells; captured results and report previews
• Drag-and-drop dashboards, KPIs, tables, charts and geographic maps
• Filters, cross-filtering and semantic-model drill-down
• Versioned ONNX model registry, optional Python training in Labs, batch prediction and drift statistics

Python notebooks and model training need your separately installed Python and the relevant packages.

• Visual app builder with tables, forms and data-bound components
• Governed writeback with a recorded write-intent history
• Published apps in a consumer-facing application catalogue
• A focused run view for the people using your app

These are QDP workspace apps. Publishing an app does not create a publicly hosted website.

• In-app assistant using Anthropic or an OpenAI-compatible endpoint, including local providers
• A localhost MCP server for external agent tools and a built-in MCP Playground
• Role-scoped Agent Portfolio with attribution, tool permissions and run history
• Build plans to review before work proceeds; recorded reasons for external-agent mutations
• Provider keys stored in the operating system keychain

Bring your own model and provider access. Agent requests and tool results may be sent to your chosen AI provider.

• QDP Cloud organisations, invitations and a shared Team Catalogue
• Opt-in sharing of contract schemas, governance metadata and quality results
• Shared workspace repositories and a CLI for headless workflows
• Microsoft Fabric Lakehouse deployment of published products, with drift monitoring
• Desktop interface in English, Spanish, French, German and Italian

Cloud sharing, shared repositories and deployment require their respective services and credentials. Team Catalogue sharing sends metadata, not table rows or Parquet files.