📊 Statisticians Workbench – Development Roadmap

Version 3.4 • 2026-08-13 • Living Document – Subject to refinement

Current Version: 0.12.0.dev (In Development) Last Release: 0.11.0 (2026-07-24) Next Release: 0.12.0

Version Status Overview

VersionMilestoneThemeStatus
0.9.01Dataset management, password policy, admin featuresReleased
0.10.02Statistical Foundation — core stats, DuckDB, recipes, sharingReleased
0.11.03AI Agents & Collaboration — chat agents, notebook import, mobile, audio I/OReleased
0.12.04Recipe Redesign & Platform Polish — master-detail, transforms, SQL engine, sharing, translationsIn Progress
0.13.05Production Readiness — audit, performance, documentation, testingPlanned

Milestone 2 – "Statistical Foundation"

✅ Complete

Delivered in: Version 0.10.0

FeatureStatus
DuckDB Storage Engine (Quack protocol)Complete
Dataset Management (Upload, Define, Browse)Complete
Recipe Framework (CRUD, execution pipeline, wizard UI)Complete
Statistics Engine (Descriptive, correlation, t-test, ANOVA, regression, chi-square)Complete
Visualisation (Bar, pie, line, scatter, histogram)Complete
User Groups (100 groups, sharing, admin management)Complete
Resource Sharing (AI, DB, Datasets, Recipes via group)Complete
Multilingual Support (Backend + Frontend i18n framework)Complete
Database Connections (PostgreSQL, MySQL via UI)Complete
Data Preparation Rules (per-dataset transforms, preview)Complete
Golden Screwdriver — MS Access via jet-serverComplete

Milestone 3 – "AI Agents & Collaboration"

✅ Complete

Delivered in: Version 0.11.0

FeatureStatus
Chat Agent — general chat with session management, CRUD, streamingComplete
AI Providers — Anthropic, Ollama, OpenAI-compatible (DeepSeek, etc.)Complete
AI Agents — Dataset Cleaner, Database Designer, Recipe AssistantComplete
AI-Augmented Database Design — Schema AI, Semantic AI (ISIC domains, code systems, column meanings), Integration AI (relationships, research capabilities, DDL proposals)Complete
Notebook Import — .ipynb parsing, DuckDB/PostgreSQL/MySQL storageComplete
Group Administration — group descriptions, admin UI, user group assignmentComplete
Example User — demo account with read-only guardrailsComplete
Audio I/O — speech-to-text + text-to-speech, per-user settingsComplete
MS SQL Server — connection, schema browsing, dataset importComplete
Mobile Support — hamburger drawer, responsive sidebars, touch targetsComplete
Copy-to-Clipboard, Demo Indicators, Feature GatesComplete

Milestone 4 – "Recipe Redesign & Platform Polish"

🔄 In Progress (v0.12.0) 12/21 complete

Theme: A complete rebuild of the recipe system with pipeline editor, dataset-level transforms, SQL execution pushdown, and platform-wide translation compliance.

FeatureStatusNotes
Recipe Master-Detail LayoutCompleteTwo-panel layout: scrollable list (left) + editor (right). Removed old wizard. Deep linking via /recipes/[id].
Three-Component Pipeline EditorCompleteDataset/Transform, Algorithm, Output components with collapsible cards, bidirectional constraint propagation.
Dataset-Level Transform ManagementCompleteTransforms editor (two-panel), 8 operation types, live preview, CRUD with recipe reference integrity.
SQL Execution EngineCompleteUniversal ops compiler (451 lines), 5 backends (DuckDB, PostgreSQL, MySQL, MSSQL, Access), capability matrix, partial pushdown with pandas fallback. 84 tests.
Recipe Sharing & CloningCompleteGroup-based share toggle, "Shared by" badge, read-only mode for non-owners, recipe forking.
DuckDB-Native StatisticsCompleteDescriptive, correlation, t-tests, ANOVA, chi-square, linear regression via DuckDB aggregates. Non-DuckDB uses scipy. 135 tests.
Preview SQL PushdownCompletePreview endpoint wired for both direct ops and rule-based modes. SQL compiled, split, executed natively.
New Settings DesignCompleteAdmin → Setup (Limits, Options, Internal, Licence). Removed Settings → Global.
Licence Management in SetupCompleteLicence page repositioned under Admin → Setup.
Auto-Create Datasets on Connection SaveCompletePostgreSQL, MySQL, MSSQL, Access. Idempotent.
Translation Compliance AuditCompleteAudited entire codebase (~80 violations). Fixed all backend and frontend violations. En.yaml regenerated (1723 keys frontend, 366 keys backend). Full translation deferred to release.
AI Recipe RestructureCompleteReplace current single-turn chat (clone of AIDataClean) with multi-turn educational concierge. 6 phases, 23 tasks. Backend endpoints implemented and working. Full verification deferred to End-to-End Verification item.
Remove Old Preparation PagesCompleteDeleted standalone /datasets/[id]/preparation routes. Redirects to /recipes. Transforms editor is the replacement.
Consolidate Duplicate ValidationCompleteExtracted 130-line validate_operations() and 4 constants into shared backend/src/routers/datasets/validation.py. Both files now call shared logic with their respective translation dicts.
Dead Code RemovalCompleteRemoved 6 dead Next.js /api/recipes route handlers (594 lines). All requests rewritten to Python backend via next.config.mjs.
Dynamic Algorithm/Output DiscoveryCompleteRecipe editor now fetches algorithm types and visualization types from backend API instead of hardcoded frontend lists. New GET /api/recipes/recipe-algorithms endpoint.
End-to-End VerificationPendingFull workflow walkthrough: upload → transform → recipe → execute → visualise.
Virtual DatasetsPlannedDuckDB federated SQL joins across uploaded CSVs, PostgreSQL, MySQL tables. AI-guided creation.
More Chart TypesCompleteScatter, area, horizontal bar, donut, histogram, distribution (sandbox + recipe pipeline).
Translation: 8-Language PassCompleteFrontend + backend translated to 8 languages via delta mode (129–341 new keys each). Imported into DB.

Milestone 5 – "Production Readiness"

📋 Planned (v0.13.0)

Theme: Making the Workbench enterprise-ready.

FeatureNotes
Code AuditSecurity audit, performance review, dead code removal, dependency audit. Pre-release gate. (moved from 4)
Performance OptimisationQuery optimisation, caching strategies
Activity Logs & Audit TrailUser activity tracking + system-wide audit logging (moved from 3)
Comprehensive TestingUnit tests, integration tests, E2E tests
User & Developer DocumentationAPI docs, architecture guides, user manuals
Backup & RecoveryAutomated backups, restore procedures
Monitoring & AlertingMetrics, dashboards, alerts

Deferred — No Target Milestone

Features deferred indefinitely. May be re-prioritised in a future roadmap revision.

FeatureReason
AI Recipe Assistant (Phase 7 — old single-turn)Replaced by AI Recipe Restructure (now in 4). Old stateless agent deprecated. Deferred 2026-07-28, superseded 2026-07-29.
Advanced Statistical TechniquesBayesian methods, time series, clustering — not required for current user base. (moved from 4)
Recipe Export/ImportMove from milestone 4 — not required for current user base.
Export VisualisationsMove from milestone 4 — not required for current user base.
Interactive VisualisationsUser interaction with charts — requires Plotly upgrade. (moved from 4)
More Database TypesSQLite, Snowflake, BigQuery — no current user demand. (moved from 5)
Recipe TemplatesPre-built recipe templates — low priority without community growth. (moved from 5)
Agent Sharing & MarketplaceEcosystem features — requires user base growth first