AI Data Cleaning logo
AI Data Cleaning
  • Features
  • How it works
  • FAQ
  • Pricing
Home/Coming soon

Coming soon

Reusable cleaning recipes need more than a saved prompt

A reliable recipe should freeze supported rules, identify the schema it was designed for, and stop when a new file no longer matches. The recipe experience is available for development review, while persistence and execution remain clearly deferred.

Start cleaningView beta access
01

What a versioned recipe is meant to preserve

The planned recipe model groups an immutable rule version with its expected columns, types, dictionary notes, safeguards, and provenance. Editing rules creates a new version instead of rewriting the historical recipe used by an earlier run.

This design keeps reuse explainable. A user should be able to see which version was selected, what changed between versions, and which frozen snapshot was ultimately confirmed for a run.

02

Schema drift must interrupt automatic reuse

Before a future recipe can run on a new file, the service will compare required columns and expected types. Missing fields, renamed columns, new columns, empty required fields, and invalid rule targets must become visible drift states rather than silent guesses.

Exact matches may proceed to review. Compatible changes can be acknowledged. Blocking drift requires a deliberate mapping or a revised recipe version. The current Worker never receives preview-only mappings or unsupported advanced rules.

  • Exact, compatible, warning, and blocking outcomes
  • Explicit source-to-target column mapping
  • Unmatched categories remain visible
  • No automatic execution after a preview match
03

Why this page says coming soon

Recipe cards, versions, drift, and mapping are currently fixtures inside the development-only preview. They do not create database rows, save user data, call an AI mapping service, or send a plan to the Worker.

The real product continues to support creating and running a plan for the current project. Public pages describe recipe reuse as planned until repository, ownership, audit, and backend validation contracts are implemented.

Current beta boundary

Real uploads, profiling, review, deterministic runs, reports and current exports are available. Recipes, credits and advanced rules remain clearly marked as coming soon.

Free during beta · no checkout

Continue exploring

Related data-cleaning guides

Reproducibility

AI suggestions can lead to reproducible cleaning when execution is separated

Auditable results

A data cleaning report should describe the run that actually happened

Categorical cleanup

Standardize categories with rules you can read and audit

AI Data Cleaning logoAI Data Cleaning

AI proposes the cleaning plan. You review and confirm before anything runs.

Support: support@tidyx.io

Product
  • Features
  • How it works
  • FAQ
  • Pricing
Company
  • Contact
Legal
  • Privacy Policy
  • Terms of Service
  • AI Usage Policy
© 2026 AI Data Cleaning. All Rights Reserved.