Roadmap
The v2.0.0 engine makes preparation rules reusable and reviewable. The current product phase brings those recipes into customer-facing workflows validated through real usage and pilots.
v0.1–0.2Data-quality foundationsShipped
- Profiling, assessment and automatic cleaning
- CSV, Excel and pandas.DataFrame inputs
- Console, HTML, PDF, Excel, CSV, JSON and Markdown reports
v0.3.0Deterministic Agentic EDAShipped
- Semantic dataset understanding and deterministic analysis planning
- Modular execution with traceable findings and actions
- HTML, JSON, Markdown and PNG EDA artifacts
v0.4.0Notebook export and human approvalShipped
- Ready-to-run deterministic notebook export
- Review checkpoints between planning and execution
- Dataset fingerprint and dependency validation
v0.5.0Grounded AI narrativesShipped
- Optional evidence-cited narratives after deterministic analysis
- Provider-neutral adapter layer
- Citation and workflow-integrity guardrails
v0.6–0.7Cleaning and preparation controlsShipped
- Dry-run cleaning, per-column rules and structured change logs
- Threshold-gated type and date conversions
- Preparation reports with warnings and before/after details
v0.8–0.9Contracts and production workflowsShipped
- Schema inference and reusable quality rules
- Explicit contract pass/fail reports
- Batch CLI, JSON/YAML configuration, JSONL logs and stable exit codes
v1.0.0Stable production APIShipped
- Unified edf.run() workflow
- Stable result/report objects and package-level errors
- Compatibility across every v0.5 public workflow
- Verified Python 3.10–3.13 release
v1.4.0Transformation-first readiness workflowsShipped
- Analysis Ready workflow with evidence, warnings and validation
- Leakage-safe ML Ready train/test preparation
- Power BI Ready single-table and multi-table model inputs
- Verified Python 3.10–3.13 release
v1.4.1Reproducible packaging maintenanceShipped
- Source packages include the sample dataset and complete test fixtures
- Documented development extra for quality, test, build and validation tools
- Supported pandas categorical type checks with no public API changes
- Verified Python 3.10–3.13 release
v2.0.0Reusable, reviewable preparationCurrent
- Versioned JSON/YAML recipes with explicit column types and mappings
- Preview/apply review plans bound to source, recipe and proposed output
- Preserve, reject and quarantine conversion policies with audit evidence
- Shared pipeline reports for Analysis Ready, ML Ready and Power BI Ready
Product betaCustomer-facing Data StudioOpen
- Public browser preview for transparent CSV inspection and export
- Browser integration for v2 recipes, Excel and Power BI workflows
- Product priorities driven by repeated customer usage and paid pilot evidence